DETAILED ACTION
This communication is in response to the Application filed on April 27, 2025.
Claims 1 - 20 are pending and have been examined.
Claims 1 and 11 are independent.
Foreign priority: October 28, 2022.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statements (IDS) submitted on August 20, 2026, October 28, 2025 and October 21, 2025 are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statements were considered by the examiner.
Drawings
The drawings filed on April 27, 2025 have been accepted and considered by the Examiner.
Double Patenting Note
The Examiner notes that previously published U.S. Patent Application Publication 2022/0012432 and 2025/0225398 were analyzed for Double Patenting. However, based on the current claim scope no Double patenting was found.
Claim Rejections - 35 USC § 103
The following is a quotation of pre-AIA 35 U.S.C. 103(a) which forms the basis for all obviousness rejections set forth in this Office action:
(a) A patent may not be obtained though the invention is not identically disclosed or described as set forth in section 102 of this title, if the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains. Patentability shall not be negatived by the manner in which the invention was made.
Claims 1 - 5 and 11 - 15 are rejected under 35 U.S.C. 103(a) as being unpatentable over Zhao et al., (CN115455982A), hereinafter referred to as Zhao, in view of Zhang et al., (CN109582798A), hereinafter referred to as Zhang, Feng et al., (U.S. Patent Application Publication 2021/0192142), hereinafter referred to as Feng.
Regarding Claim 1, Zhao teaches:
1. A human-computer dialogue method implemented by an electronic device, comprising:
obtaining second historical dialogue data, current dialogue environment data, and second input dialogue data from a user; [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048]
determining a second semantic keyword based on the second historical dialogue data, the current dialogue environment data, and the second input dialogue data; and [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048]
generating and outputting, based on the second semantic keyword and a preset-type knowledge base, second feedback dialogue data corresponding to the second input dialogue data, [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined (i.e., the claimed “second feedback dialogue data corresponding to the second input dialogue data”). By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048]
wherein the preset-type knowledge base comprises a knowledge graph, text knowledge, and external knowledge other than the knowledge graph and the text knowledge. [Zhao, “according to a pre-constructed dialogue knowledge graph,” Par. n0048; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term, the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term, an information push operation can be performed. The information push operation can call an external knowledge base. For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079]
While Zhao teaches the limitations, Zhang also teaches:
obtaining second historical dialogue data, current dialogue environment data, and second input dialogue data from a user; [Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” Par. 0064; “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081]
determining a second semantic keyword based on the second historical dialogue data, the current dialogue environment data, and the second input dialogue data; and [Zhang, “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178]
While Zhao in view of Zhang teaches the limitations, Feng also teaches:
obtaining second historical dialogue data, current dialogue environment data, and second input dialogue data from a user; [Feng, “For example, a tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “current dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” Par. 0078; Referring to Specification Par. 00187 of the instant Application, “current dialogue environment data” may be obtained “based on a sensor disposed in the application apparatus, and sensors in different application scenarios may be different. For example, in a scenario of a dialogue between an indoor user and a smart speaker, the sensor may be an image capture apparatus. For another example, in a scenario of a dialogue between an outdoor user and a vehicle-mounted terminal, the sensor may be a speed sensor.”]
determining a second semantic keyword based on the second historical dialogue data, the current dialogue environment data, and the second input dialogue data; and [Feng, “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018]
Zhao, Zhang and Feng pertain to semantic understanding dialogue systems and are analogous to the instant application. Accordingly, it would have been obvious to one of ordinary skill in the semantic understanding dialogue systems art to modify Zhao’s teachings of “dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data from a user”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management” (Zhao, Par. n0048) with the teachings of “multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” (Zhang, Par. 0064) taught by Zhang, and the teachings of “tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “current dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” (Feng, Par. 0078) taught by Feng in order “create a dialogue experience” and increase “the accuracy as multiple rounds of dialogue proceed” (Zhang, Par. 0064) and to solve the problem of “clarifying a relationship between the knowledge nodes, and an event that actually occurred, and therefore, a problem arises that a semantic understanding result is inaccurate.” (Feng, Par. 0005).
Regarding Claim 2, Zhao in view of Zhang and Feng has been discussed above. The combination further teaches:
wherein the generating and outputting, based on the second semantic keyword and the preset-type knowledge base, second feedback dialogue data corresponding to the second input dialogue data comprises: [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined (i.e., the claimed “second feedback dialogue data corresponding to the second input dialogue data”). By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048]
separately obtaining, from the external knowledge, the knowledge graph, and the text knowledge, knowledge corresponding to the second semantic keyword; and [Zhao, “according to a pre-constructed dialogue knowledge graph,” Par. n0048; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term, the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term, an information push operation can be performed. The information push operation can call an external knowledge base. For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; Zhang, “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178]
generating and outputting the second feedback dialogue data based on the obtained knowledge corresponding to the second semantic keyword. [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined (i.e., the claimed “second feedback dialogue data corresponding to the second input dialogue data”). By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048]
Regarding Claim 3, Zhao in view of Zhang and Feng has been discussed above. The combination further teaches:
wherein after the separately obtaining, from the external knowledge, the knowledge graph, and the text knowledge, knowledge corresponding to the second semantic keyword, the method further comprises: [Zhao, “according to a pre-constructed dialogue knowledge graph,” Par. n0048; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term, the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term, an information push operation can be performed. The information push operation can call an external knowledge base. For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; Zhang, “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178]
retrieving, from at least one of other types of knowledge based on knowledge that is obtained from first-type knowledge and that corresponds to the second semantic keyword, knowledge associated with the knowledge that is obtained from the first-type knowledge and that corresponds to the second semantic keyword, [Zhao, “according to a pre-constructed dialogue knowledge graph,” Par. n0048; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term, the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term, an information push operation can be performed. The information push operation can call an external knowledge base. For example, if the slot value is a location (i.e., the claimed “first type of knowledge”), a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate (i.e., the claimed “first type of knowledge”), a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; Zhang, “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178]
wherein the knowledge that is in the at least one of other types of knowledge and that corresponds to the second semantic keyword comprises the retrieved knowledge associated with the knowledge that is obtained from the first-type knowledge and that corresponds to the second semantic keyword, the first-type knowledge is any one of the external knowledge, the knowledge graph, or the text knowledge, and [Zhao, “according to a pre-constructed dialogue knowledge graph,” Par. n0048; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term, the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term, an information push operation can be performed. The information push operation can call an external knowledge base. For example, if the slot value is a location (i.e., the claimed “first type of knowledge”), a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate (i.e., the claimed “first type of knowledge”), a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; Zhang, “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178]
the at least one of other types of knowledge is type knowledge other than the first-type knowledge in the external knowledge, the knowledge graph, and the text knowledge. [Zhao, “according to a pre-constructed dialogue knowledge graph,” Par. n0048; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term, the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term, an information push operation can be performed. The information push operation can call an external knowledge base. For example, if the slot value is a location (i.e., the claimed “first type of knowledge”), a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather (i.e., the claimed “other types of knowledge is type knowledge other than the first-type knowledge”) for that location (i.e., the claimed “first type of knowledge”), and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map (i.e., the claimed “other types of knowledge is type knowledge other than the first-type knowledge”) for that location (i.e., the claimed “first type of knowledge”). If the domain is real estate (i.e., the claimed “first type of knowledge”), a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices (i.e., the claimed “other types of knowledge is type knowledge other than the first-type knowledge”) for that location (i.e., the claimed “first type of knowledge”).” Par. n0079; Zhang, “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; Zhang, “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178]
Regarding Claim 4, Zhao in view of Zhang and Feng has been discussed above. The combination further teaches:
wherein the generating and outputting the second feedback dialogue data based on the obtained knowledge corresponding to the second semantic keyword comprises: [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined (i.e., the claimed “second feedback dialogue data corresponding to the second input dialogue data”). By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048]
separately encoding the obtained knowledge corresponding to the second semantic keyword, to obtain corresponding target feature vectors; and [Zhao, “Based on the pre-constructed dialogue knowledge graph, the domain information, intent information and semantic slot information in the first semantic information and the historical dialogue semantic information are vectorized to obtain the first semantic vector and the historical dialogue semantic information vector.” Par. n001; “Based on the first semantic information vector (i.e., the claimed “corresponding target feature vectors”) and the historical dialog semantic information vector (i.e., the claimed “corresponding target feature vectors”), the value of each vector bit of the first semantic information vector is re-determined to obtain the second semantic information vector (i.e., the claimed “corresponding target feature vectors”).” Par. n0011]
generating and outputting the second feedback dialogue data based on the corresponding target feature vectors. [Zhao, “The dialogue strategy determination module is used to determine the dialogue prompt information (i.e., the claimed “second feedback dialogue data”) corresponding to the second semantic information based on the second semantic information.” Par. n0043; “Based on the pre-constructed dialogue knowledge graph, the domain information, intent information and semantic slot information in the first semantic information and the historical dialogue semantic information are vectorized to obtain the first semantic vector and the historical dialogue semantic information vector.” Par. n0010; “Optionally, the second semantic information is determined based on the second semantic information vector (i.e., the claimed “corresponding target feature vector”) and the dialogue knowledge graph,” Par. n0016]
Regarding Claim 5, Zhao in view of Zhang and Feng has been discussed above. The combination further teaches:
wherein the generating and outputting the second feedback dialogue data based on the corresponding target feature vectors comprises: [Zhao, “The dialogue strategy determination module is used to determine the dialogue prompt information (i.e., the claimed “second feedback dialogue data”) corresponding to the second semantic information based on the second semantic information.” Par. n0043; “Based on the pre-constructed dialogue knowledge graph, the domain information, intent information and semantic slot information in the first semantic information and the historical dialogue semantic information are vectorized to obtain the first semantic vector and the historical dialogue semantic information vector.” Par. n0010; “Optionally, the second semantic information is determined based on the second semantic information vector (i.e., the claimed “corresponding target feature vector”) and the dialogue knowledge graph,” Par. n0016]
fusing the corresponding target feature vectors to obtain a target fusion feature vector; and [Feng, “vector representation of the semantic understanding result,” Par. 0030; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results (i.e., the claimed “target fusion feature vector”) corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018; Zhao, “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together (i.e., the claimed “target fusion feature vector”), which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhao, “Based on the pre-constructed dialogue knowledge graph, the domain information, intent information and semantic slot information in the first semantic information and the historical dialogue semantic information are vectorized to obtain the first semantic vector and the historical dialogue semantic information vector.” Par. n0010; “Based on the first semantic information vector (i.e., the claimed “corresponding target feature vectors”) and the historical dialog semantic information vector (i.e., the claimed “corresponding target feature vectors”), the value of each vector bit of the first semantic information vector is re-determined to obtain the second semantic information vector (i.e., the claimed “corresponding target feature vectors”).” Par. n0011]
generating and outputting the second feedback dialogue data based on the target fusion feature vector. [Feng, “vector representation of the semantic understanding result,” Par. 0030; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018; Zhao, “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information (i.e., the claimed “second feedback dialogue data”) corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhao, “Based on the pre-constructed dialogue knowledge graph, the domain information, intent information and semantic slot information in the first semantic information and the historical dialogue semantic information are vectorized to obtain the first semantic vector and the historical dialogue semantic information vector.” Par. n0010; “Based on the first semantic information vector (i.e., the claimed “corresponding target feature vectors”) and the historical dialog semantic information vector (i.e., the claimed “corresponding target feature vectors”), the value of each vector bit of the first semantic information vector is re-determined to obtain the second semantic information vector (i.e., the claimed “corresponding target feature vectors”).” Par. n0011]
Regarding Claim 11, Zhao in view of Zhang and Feng has been discussed above. The combination further teaches:
11. A dialogue network model training method implemented by an electronic device, comprising:
obtaining a second sample data set, wherein the second sample data set comprises second sample input dialogue data input by a user, second sample historical dialogue data, and first sample dialogue environment data corresponding to the second sample input dialogue data; [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “first sample dialogue environment data”), delivery address (i.e., the claimed “first sample dialogue environment data”), and delivery time (i.e., the claimed “first sample dialogue environment data”). The current dialog text (i.e., the claimed “second sample input dialogue data”) only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second sample historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “first sample dialogue environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second sample input dialogue data”)” Par. 0064; “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178; Feng, “For example, a tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “first sample dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” Par. 0078; Referring to Specification Par. 00187 of the instant Application, “current dialogue environment data” may be obtained “based on a sensor disposed in the application apparatus, and sensors in different application scenarios may be different. For example, in a scenario of a dialogue between an indoor user and a smart speaker, the sensor may be an image capture apparatus. For another example, in a scenario of a dialogue between an outdoor user and a vehicle-mounted terminal, the sensor may be a speed sensor.”; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018]
generating a second sample semantic keyword based on the second sample data set; and [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “first sample dialogue environment data”), delivery address (i.e., the claimed “first sample dialogue environment data”), and delivery time (i.e., the claimed “first sample dialogue environment data”). The current dialog text (i.e., the claimed “second sample input dialogue data”) only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second sample semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second sample historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “first sample dialogue environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second sample input dialogue data”)” Par. 0064; “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178; Feng, “For example, a tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “first sample dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” Par. 0078; Referring to Specification Par. 00187 of the instant Application, “current dialogue environment data” may be obtained “based on a sensor disposed in the application apparatus, and sensors in different application scenarios may be different. For example, in a scenario of a dialogue between an indoor user and a smart speaker, the sensor may be an image capture apparatus. For another example, in a scenario of a dialogue between an outdoor user and a vehicle-mounted terminal, the sensor may be a speed sensor.”; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018]
obtaining, through training based on the second sample semantic keyword and a preset-type knowledge base, a dialogue network model that has a capability of predicting feedback dialogue data corresponding to input dialogue data, [Zhao, “In this embodiment, the loss function used during training of the semantic information (i.e., the claimed “second sample semantic keyword”) classification model is:” Par. n0105; “Furthermore, this embodiment can also pre-train a natural language understanding model (i.e., the claimed “dialogue network model”), input the current dialogue text or the concatenated text of the current dialogue text and the historical dialogue text (i.e., the claimed “input dialogue data”) in the natural language understanding model, use the natural language understanding model to analyze and predict the text (i.e., the claimed “predicting feedback dialogue data”),” Par. n0070; Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “first sample dialogue environment data”), delivery address (i.e., the claimed “first sample dialogue environment data”), and delivery time (i.e., the claimed “first sample dialogue environment data”). The current dialog text (i.e., the claimed “second sample input dialogue data”) only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second sample semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second sample historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “first sample dialogue environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second sample input dialogue data”)” Par. 0064; “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178; Feng, “For example, a tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “first sample dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” Par. 0078; Referring to Specification Par. 00187 of the instant Application, “current dialogue environment data” may be obtained “based on a sensor disposed in the application apparatus, and sensors in different application scenarios may be different. For example, in a scenario of a dialogue between an indoor user and a smart speaker, the sensor may be an image capture apparatus. For another example, in a scenario of a dialogue between an outdoor user and a vehicle-mounted terminal, the sensor may be a speed sensor.”; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018]
wherein the preset-type knowledge base comprises a knowledge graph, text knowledge, and external knowledge other than the knowledge graph and the text knowledge. [Zhao, “according to a pre-constructed dialogue knowledge graph,” Par. n0048; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term, the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term, an information push operation can be performed. The information push operation can call an external knowledge base. For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079]
Regarding Claim 12, Zhao in view of Zhang and Feng has been discussed above. The combination further teaches:
wherein the obtaining, through training based on the second sample semantic keyword and the preset-type knowledge base, the dialogue network model that has a capability of predicting feedback dialogue data corresponding to input dialogue data comprises: [Zhao, “In this embodiment, the loss function used during training of the semantic information (i.e., the claimed “second sample semantic keyword”) classification model is:” Par. n0105; “Furthermore, this embodiment can also pre-train a natural language understanding model (i.e., the claimed “dialogue network model”), input the current dialogue text or the concatenated text of the current dialogue text and the historical dialogue text (i.e., the claimed “input dialogue data”) in the natural language understanding model, use the natural language understanding model to analyze and predict the text (i.e., the claimed “predicting feedback dialogue data”),” Par. n0070; Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “first sample dialogue environment data”), delivery address (i.e., the claimed “first sample dialogue environment data”), and delivery time (i.e., the claimed “first sample dialogue environment data”). The current dialog text (i.e., the claimed “second sample input dialogue data”) only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second sample semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second sample historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “first sample dialogue environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second sample input dialogue data”)” Par. 0064; “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178; Feng, “For example, a tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “first sample dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” Par. 0078; Referring to Specification Par. 00187 of the instant Application, “current dialogue environment data” may be obtained “based on a sensor disposed in the application apparatus, and sensors in different application scenarios may be different. For example, in a scenario of a dialogue between an indoor user and a smart speaker, the sensor may be an image capture apparatus. For another example, in a scenario of a dialogue between an outdoor user and a vehicle-mounted terminal, the sensor may be a speed sensor.”; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018]
separately obtaining, from the external knowledge, the knowledge graph, and the text knowledge, knowledge corresponding to the second sample semantic keyword; and [Zhao, “according to a pre-constructed dialogue knowledge graph,” Par. n0048; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term, the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term, an information push operation can be performed. The information push operation can call an external knowledge base. For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; Zhang, “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178]
training the dialogue network model through training based on the obtained knowledge corresponding to the second sample semantic keyword. [Zhao, “In this embodiment, the loss function used during training of the semantic information (i.e., the claimed “second sample semantic keyword”) classification model is:” Par. n0105; “Furthermore, this embodiment can also pre-train a natural language understanding model (i.e., the claimed “dialogue network model”), input the current dialogue text or the concatenated text of the current dialogue text and the historical dialogue text (i.e., the claimed “input dialogue data”) in the natural language understanding model, use the natural language understanding model to analyze and predict the text (i.e., the claimed “predicting feedback dialogue data”),” Par. n0070; Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “first sample dialogue environment data”), delivery address (i.e., the claimed “first sample dialogue environment data”), and delivery time (i.e., the claimed “first sample dialogue environment data”). The current dialog text (i.e., the claimed “second sample input dialogue data”) only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second sample semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second sample historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “first sample dialogue environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second sample input dialogue data”)” Par. 0064; “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178; Feng, “For example, a tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “first sample dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” Par. 0078; Referring to Specification Par. 00187 of the instant Application, “current dialogue environment data” may be obtained “based on a sensor disposed in the application apparatus, and sensors in different application scenarios may be different. For example, in a scenario of a dialogue between an indoor user and a smart speaker, the sensor may be an image capture apparatus. For another example, in a scenario of a dialogue between an outdoor user and a vehicle-mounted terminal, the sensor may be a speed sensor.”; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018]
Regarding Claim 13, Zhao in view of Zhang and Feng has been discussed above. The combination further teaches:
wherein after the separately obtaining, from the external knowledge, the knowledge graph, and the text knowledge, knowledge corresponding to the second sample semantic keyword, the method further comprises: [Zhao, “according to a pre-constructed dialogue knowledge graph,” Par. n0048; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term, the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term, an information push operation can be performed. The information push operation can call an external knowledge base. For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; Zhang, “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178]
retrieving, from at least one of other types of knowledge based on knowledge that is obtained from first-type knowledge and that corresponds to the second sample semantic keyword, knowledge associated with the knowledge that is obtained from the first-type knowledge and that corresponds to the second sample semantic keyword, wherein the knowledge that is in the at least one of other types of knowledge and that corresponds to the second sample semantic keyword comprises the retrieved knowledge associated with the knowledge that is obtained from the first-type knowledge and that corresponds to the second sample semantic keyword, the first-type knowledge is any one of the external knowledge, the knowledge graph, or the text knowledge, and the at least one of other types of knowledge is type knowledge other than the first-type knowledge in the external knowledge, the knowledge graph, and the text knowledge. [Zhao, “according to a pre-constructed dialogue knowledge graph,” Par. n0048; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term, the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term, an information push operation can be performed. The information push operation can call an external knowledge base. For example, if the slot value is a location (i.e., the claimed “first type of knowledge”), a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate (i.e., the claimed “first type of knowledge”), a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; Zhang, “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second sample semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second sample historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “first sample dialogue environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048]
Regarding Claim 14, Zhao in view of Zhang and Feng has been discussed above. The combination further teaches:
wherein the obtaining the dialogue network model through training based on the obtained knowledge corresponding to the second sample semantic keyword comprises: [Zhao, “In this embodiment, the loss function used during training of the semantic information (i.e., the claimed “second sample semantic keyword”) classification model is:” Par. n0105; “Furthermore, this embodiment can also pre-train a natural language understanding model (i.e., the claimed “dialogue network model”), input the current dialogue text or the concatenated text of the current dialogue text and the historical dialogue text (i.e., the claimed “input dialogue data”) in the natural language understanding model, use the natural language understanding model to analyze and predict the text (i.e., the claimed “predicting feedback dialogue data”),” Par. n0070; Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “first sample dialogue environment data”), delivery address (i.e., the claimed “first sample dialogue environment data”), and delivery time (i.e., the claimed “first sample dialogue environment data”). The current dialog text (i.e., the claimed “second sample input dialogue data”) only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second sample semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second sample historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “first sample dialogue environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second sample input dialogue data”)” Par. 0064; “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178; Feng, “For example, a tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “first sample dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” Par. 0078; Referring to Specification Par. 00187 of the instant Application, “current dialogue environment data” may be obtained “based on a sensor disposed in the application apparatus, and sensors in different application scenarios may be different. For example, in a scenario of a dialogue between an indoor user and a smart speaker, the sensor may be an image capture apparatus. For another example, in a scenario of a dialogue between an outdoor user and a vehicle-mounted terminal, the sensor may be a speed sensor.”; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018]
separately encoding the obtained knowledge corresponding to the second sample semantic keyword, to obtain corresponding feature vectors; and [Zhao, “Based on the first semantic information vector (i.e., the claimed “corresponding target feature vectors”) and the historical dialog semantic information vector (i.e., the claimed “corresponding target feature vectors”), the value of each vector bit of the first semantic information vector is re-determined to obtain the second semantic information vector (i.e., the claimed “corresponding target feature vectors”).” Par. n0011; “Based on the text vectors corresponding to all text characters (i.e., the claimed “text knowledge”), a classification embedding feature vector corresponding to the current dialogue text is calculated.” Par. n0031]
obtaining the dialogue network model through training based on the corresponding feature vectors. [Zhao, “In this embodiment, the loss function used during training of the semantic information (i.e., the claimed “second sample semantic keyword”) classification model is:” Par. n0105; “Furthermore, this embodiment can also pre-train a natural language understanding model (i.e., the claimed “dialogue network model”), input the current dialogue text or the concatenated text of the current dialogue text and the historical dialogue text (i.e., the claimed “input dialogue data”) in the natural language understanding model, use the natural language understanding model to analyze and predict the text (i.e., the claimed “predicting feedback dialogue data”),” Par. n0070; Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “first sample dialogue environment data”), delivery address (i.e., the claimed “first sample dialogue environment data”), and delivery time (i.e., the claimed “first sample dialogue environment data”). The current dialog text (i.e., the claimed “second sample input dialogue data”) only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second sample semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second sample historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “first sample dialogue environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second sample input dialogue data”)” Par. 0064; “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; “Using the relationship information between the first instance ontology and the second instance ontology, search for the second semantic node in at one set of structured graph knowledge.” Par. 0178; Feng, “For example, a tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “first sample dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” Par. 0078; Referring to Specification Par. 00187 of the instant Application, “current dialogue environment data” may be obtained “based on a sensor disposed in the application apparatus, and sensors in different application scenarios may be different. For example, in a scenario of a dialogue between an indoor user and a smart speaker, the sensor may be an image capture apparatus. For another example, in a scenario of a dialogue between an outdoor user and a vehicle-mounted terminal, the sensor may be a speed sensor.”; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018; Zhao, “Based on the first semantic information vector (i.e., the claimed “corresponding target feature vectors”) and the historical dialog semantic information vector (i.e., the claimed “corresponding target feature vectors”), the value of each vector bit of the first semantic information vector is re-determined to obtain the second semantic information vector (i.e., the claimed “corresponding target feature vectors”).” Par. n0011; “Based on the text vectors corresponding to all text characters (i.e., the claimed “text knowledge”), a classification embedding feature vector corresponding to the current dialogue text is calculated.” Par. n0031]
Regarding Claim 15, Zhao in view of Zhang, and Feng has been discussed above. The combination further teaches:
wherein the obtaining the dialogue network model through training based on the corresponding feature vectors comprises:
fusing the corresponding feature vectors to obtain a fusion feature vector; [Feng, “vector representation of the semantic understanding result,” Par. 0030; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results (i.e., the claimed “target fusion feature vector”) corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018; Zhao, “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together (i.e., the claimed “target fusion feature vector”), which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhao, “Based on the pre-constructed dialogue knowledge graph, the domain information, intent information and semantic slot information in the first semantic information and the historical dialogue semantic information are vectorized to obtain the first semantic vector and the historical dialogue semantic information vector.” Par. n0010; “Based on the first semantic information vector (i.e., the claimed “corresponding target feature vectors”) and the historical dialog semantic information vector (i.e., the claimed “corresponding target feature vectors”), the value of each vector bit of the first semantic information vector is re-determined to obtain the second semantic information vector (i.e., the claimed “corresponding target feature vectors”).” Par. n0011]
inputting the fusion feature vector into a fourth neural network model, to output second predicted dialogue data fed back to the user; and [Feng, “vector representation of the semantic understanding result,” Par. 0030; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results (i.e., the claimed “target fusion feature vector”) corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018; Zhao, “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together (i.e., the claimed “target fusion feature vector”), which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhao, “Based on the pre-constructed dialogue knowledge graph, the domain information, intent information and semantic slot information in the first semantic information and the historical dialogue semantic information are vectorized to obtain the first semantic vector and the historical dialogue semantic information vector.” Par. n0010; “Based on the first semantic information vector (i.e., the claimed “corresponding target feature vectors”) and the historical dialog semantic information vector (i.e., the claimed “corresponding target feature vectors”), the value of each vector bit of the first semantic information vector is re-determined to obtain the second semantic information vector (i.e., the claimed “corresponding target feature vectors”).” Par. n0011; Zhang, “deep neural network model, such as a recurrent neural network (RNN) model (i.e., the claimed “fourth neural network model based on the second predicted dialogue data fed back to the user”), Par. 0183; Zhao, “feedforward neural network (i.e., the claimed “fourth neural network model based on the second predicted dialogue data fed back to the user”), Par. n0146]
adjusting a parameter of the fourth neural network model based on the second predicted dialogue data fed back to the user and second pre-annotated dialogue data fed back to the user, to obtain the dialogue network model. [Zhao, “complete sentences in the identified text are sequence-annotated (i.e., the claimed “second pre-annotated dialogue data”),” Par. n0126; “feedforward neural network (i.e., the claimed “fourth neural network model based on the second predicted dialogue data fed back to the user”), Par. n0146; Zhang, “deep neural network model, such as a recurrent neural network (RNN) model (i.e., the claimed “fourth neural network model based on the second predicted dialogue data fed back to the user”), Par. 0183]
Claims 6 - 8, and 16 - 18 are rejected under 35 U.S.C. 103(a) as being unpatentable over Zhao in view of Zhang and Feng as applied in claim 5 above, and in further view of Su et al., (CN114817494A), hereinafter referred to as Su.
Regarding Claim 6, Zhao in view of Zhang and Feng has been discussed above. The combination further teaches:
wherein the target fusion feature vector is obtained by inputting the corresponding target feature vectors into a cross-attention network model obtained through pre-training. [Feng, “vector representation of the semantic understanding result,” Par. 0030; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results (i.e., the claimed “target fusion feature vector”) corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018; Zhao, “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together (i.e., the claimed “target fusion feature vector”), which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhao, “Based on the pre-constructed dialogue knowledge graph, the domain information, intent information and semantic slot information in the first semantic information and the historical dialogue semantic information are vectorized to obtain the first semantic vector and the historical dialogue semantic information vector.” Par. n0010; “Based on the first semantic information vector (i.e., the claimed “corresponding target feature vectors”) and the historical dialog semantic information vector (i.e., the claimed “corresponding target feature vectors”), the value of each vector bit of the first semantic information vector is re-determined to obtain the second semantic information vector (i.e., the claimed “corresponding target feature vectors”).” Par. n0011]
The combination fails to teach cross-attention network model obtained through pre-training.
However, Su teaches:
wherein the target fusion feature vector is obtained by inputting the corresponding target feature vectors into a cross-attention network model obtained through pre-training. [Su, “This method is based on the pre-trained language model BERT, which can effectively utilize the powerful semantic capture capabilities learned by BERT during pre-training on large general corpora. Furthermore, by continuing pre-training on the target corpus, the method uses a domain-adaptive BERT that has learned semantic representations specific to the target corpus as the encoder. The method also includes a base cross-attention interaction network (i.e., the claimed “cross-attention network model”) to alleviate the accuracy degradation caused by separate encoding.” Par. n0005]
Zhao, Zhang, Feng and Su pertain to dialogue systems and are analogous to the instant application. Accordingly, it would have been obvious to one of ordinary skill in the dialogue systems art to modify Zhao’s teachings of “dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data from a user”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management” (Zhao, Par. n0048) with the teachings of “multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” (Zhang, Par. 0064) taught by Zhang, the teachings of “tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “current dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” (Feng, Par. 0078) taught by Feng and the teachings of cross-attention interaction network (i.e., the claimed “cross-attention network model”) and “pre training” (Su, Par. n0005) taught by Su in order “create a dialogue experience” and increase “the accuracy as multiple rounds of dialogue proceed” (Zhang, Par. 0064) and to solve the problem of “clarifying a relationship between the knowledge nodes, and an event that actually occurred, and therefore, a problem arises that a semantic understanding result is inaccurate.” (Feng, Par. 0005), and “retrieve the best response from several candidate responses” (Su, Par. n0010).
Regarding Claim 7, Zhao in view of Zhang, Feng and Su has been discussed above. The combination further teaches:
wherein the generating and outputting the second feedback dialogue data based on the obtained knowledge corresponding to the second semantic keyword comprises: [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined (i.e., the claimed “second feedback dialogue data corresponding to the second input dialogue data”). By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048]
performing redundancy removal on the obtained knowledge corresponding to the second semantic keyword, to obtain knowledge after redundancy removal; and [Zhao, “redundant text in the identified text (i.e., the claimed “obtained knowledge”) is removed (i.e., the claimed “redundancy removal”),” Par. n0024]
generating and outputting the second feedback dialogue data based on the knowledge obtained after redundancy removal. [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined (i.e., the claimed “second feedback dialogue data corresponding to the second input dialogue data”). By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhao, “redundant text in the identified text (i.e., the claimed “obtained knowledge”) is removed (i.e., the claimed “redundancy removal”),” Par. n0024]
Regarding Claim 8, Zhao in view of Zhang, Feng and Su has been discussed above. The combination further teaches:
wherein determining, based on the second historical dialogue data, the current dialogue environment data, and the second input dialogue data, [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” Par. 0064; “three different types of knowledge: semi-structured question-answer pair knowledge, structured graph knowledge, and unstructured text knowledge.” Par. 0081; Feng, “For example, a tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “current dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” Par. 0078; Referring to Specification Par. 00187 of the instant Application, “current dialogue environment data” may be obtained “based on a sensor disposed in the application apparatus, and sensors in different application scenarios may be different. For example, in a scenario of a dialogue between an indoor user and a smart speaker, the sensor may be an image capture apparatus. For another example, in a scenario of a dialogue between an outdoor user and a vehicle-mounted terminal, the sensor may be a speed sensor.”]
a third classification result indicating whether the external knowledge needs to be introduced, [Zhao, “The first semantic information vector and the historical dialogue semantic information vector are input into a pre-trained semantic information classification model,” Par. n0014; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term (i.e., the claimed “classification result”), the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term (i.e., the claimed “third classification result”), an information push operation can be performed. The information push operation can call an external knowledge base (i.e., the claimed “whether the external knowledge needs to be introduced”). For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” Par. 0064]
wherein the generating and outputting, based on the second semantic keyword and a preset- type knowledge base, second feedback dialogue data corresponding to the second input dialogue data comprises: [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined (i.e., the claimed “second feedback dialogue data corresponding to the second input dialogue data”). By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048]
when the third classification result indicates that the external knowledge needs to be introduced, [Zhao, “The first semantic information vector and the historical dialogue semantic information vector are input into a pre-trained semantic information classification model,” Par. n0014; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term (i.e., the claimed “classification result”), the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term (i.e., the claimed “third classification result”), an information push operation can be performed. The information push operation can call an external knowledge base (i.e., the claimed “whether the external knowledge needs to be introduced”). For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” Par. 0064]
generating and outputting, based on the second semantic keyword and the preset- type knowledge base, the second feedback dialogue data corresponding to the second input dialogue data. [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined (i.e., the claimed “second feedback dialogue data corresponding to the second input dialogue data”). By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048]
Regarding Claim 16, Zhao in view of Zhang, Feng and Su has been discussed above. The combination further teaches:
wherein the fusion feature vector is obtained by inputting the corresponding feature vectors into a cross-attention network model obtained through pre-training. [Feng, “vector representation of the semantic understanding result,” Par. 0030; “the second semantic understanding result (i.e., the claimed “second semantic keyword”) is obtained by fusing multiple single-modal semantic understanding results (i.e., the claimed “target fusion feature vector”) corresponding to the multimodal knowledge nodes according to a preset fusion method.” Par. 0018; Zhao, “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph, the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together (i.e., the claimed “target fusion feature vector”), which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhao, “Based on the pre-constructed dialogue knowledge graph, the domain information, intent information and semantic slot information in the first semantic information and the historical dialogue semantic information are vectorized to obtain the first semantic vector and the historical dialogue semantic information vector.” Par. n0010; “Based on the first semantic information vector (i.e., the claimed “corresponding target feature vectors”) and the historical dialog semantic information vector (i.e., the claimed “corresponding target feature vectors”), the value of each vector bit of the first semantic information vector is re-determined to obtain the second semantic information vector (i.e., the claimed “corresponding target feature vectors”).” Par. n0011; Su, “This method is based on the pre-trained language model BERT, which can effectively utilize the powerful semantic capture capabilities learned by BERT during pre-training on large general corpora. Furthermore, by continuing pre-training on the target corpus, the method uses a domain-adaptive BERT that has learned semantic representations specific to the target corpus as the encoder. The method also includes a base cross-attention interaction network (i.e., the claimed “cross-attention network model”) to alleviate the accuracy degradation caused by separate encoding.” Par. n0005]
Regarding Claim 17, Zhao in view of Zhang, Feng and Su has been discussed above. The combination further teaches:
wherein the obtaining the dialogue network model through training based on the obtained knowledge corresponding to the second sample semantic keyword comprises:
performing redundancy removal on the obtained knowledge corresponding to the second sample semantic keyword, to obtain knowledge after redundancy removal; and [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined (i.e., the claimed “second feedback dialogue data corresponding to the second input dialogue data”). By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhao, “redundant text in the identified text (i.e., the claimed “obtained knowledge”) is removed (i.e., the claimed “redundancy removal”),” Par. n0024]
obtaining the dialogue network model through training based on the knowledge obtained after redundancy removal. [Zhao, “If the intent is set to query, the semantic slots include: tracking number, current location (i.e., the claimed “current environment data”), delivery address (i.e., the claimed “current environment data”), and delivery time (i.e., the claimed “current environment data”). The current dialog text only contains text with a semantic slot of odd number.“ Par. n0068; Referring to Specification Par. 00128 of the instant Application, “environment” refers to “describe an object in a dialogue scenario, for example, data used to describe a venue in the dialogue scenario, data used to describe a status of the user in the dialogue scenario, or data used to describe an emotion of the user in the dialogue scenario.”; “The dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information (i.e., the claimed “second semantic keyword”), wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined (i.e., the claimed “second feedback dialogue data corresponding to the second input dialogue data”). By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management.” Par. n0048; Zhao, “redundant text in the identified text (i.e., the claimed “obtained knowledge”) is removed (i.e., the claimed “redundancy removal”),” Par. n0024; Zhao, “In this embodiment, the loss function used during training of the semantic information (i.e., the claimed “second sample semantic keyword”) classification model is:” Par. n0105; “Furthermore, this embodiment can also pre-train a natural language understanding model (i.e., the claimed “dialogue network model”), input the current dialogue text or the concatenated text of the current dialogue text and the historical dialogue text (i.e., the claimed “input dialogue data”) in the natural language understanding model, use the natural language understanding model to analyze and predict the text (i.e., the claimed “predicting feedback dialogue data”),” Par. n0070]
Regarding Claim 18, Zhao in view of Zhang, Feng and Su has been discussed above. The combination further teaches:
determining, based on the second sample data set, a first classification result indicating whether the external knowledge needs to be introduced, wherein the obtaining, through training based on the second sample semantic keyword and the preset-type knowledge base, the dialogue network model that has a capability of predicting feedback dialogue data corresponding to input dialogue data comprises: [Zhao, “The first semantic information vector and the historical dialogue semantic information vector are input into a pre-trained semantic information classification model,” Par. n0014; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term (i.e., the claimed “classification result”), the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term (i.e., the claimed “first classification result”), an information push operation can be performed. The information push operation can call an external knowledge base (i.e., the claimed “whether the external knowledge needs to be introduced”). For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” Par. 0064]
when the first classification result indicates that the external knowledge needs to be introduced, obtaining the dialogue network model through training based on the second sample semantic keyword and the preset-type knowledge base. [Zhao, “The first semantic information vector and the historical dialogue semantic information vector are input into a pre-trained semantic information classification model,” Par. n0014; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term (i.e., the claimed “classification result”), the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term (i.e., the claimed “third classification result”), an information push operation can be performed. The information push operation can call an external knowledge base (i.e., the claimed “whether the external knowledge needs to be introduced”). For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” Par. 0064; Zhao, “In this embodiment, the loss function used during training of the semantic information (i.e., the claimed “second sample semantic keyword”) classification model is:” Par. n0105; “Furthermore, this embodiment can also pre-train a natural language understanding model (i.e., the claimed “dialogue network model”), input the current dialogue text or the concatenated text of the current dialogue text and the historical dialogue text (i.e., the claimed “input dialogue data”) in the natural language understanding model, use the natural language understanding model to analyze and predict the text (i.e., the claimed “predicting feedback dialogue data”),” Par. n0070]
Claims 9 - 10 and 19 - 20 are rejected under 35 U.S.C. 103(a) as being unpatentable over Zhao in view of Zhang, Feng and Su as applied in claim 8 above, and in further view of Xu et al., (U.S. Patent Application Publication 2022/0171947), hereinafter referred to as Xu.
Regarding Claim 9, Zhao in view of Zhang, Feng and Su has been discussed above. The combination further teaches:
wherein the third classification result is obtained by inputting the second historical dialogue data, the current dialogue environment data, and the second input dialogue data into a pre-trained binary classification network model, and [Zhao, “The first semantic information vector and the historical dialogue semantic information vector are input into a pre-trained semantic information classification model,” Par. n0014; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term (i.e., the claimed “classification result”), the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term (i.e., the claimed “third classification result”), an information push operation can be performed. The information push operation can call an external knowledge base (i.e., the claimed “whether the external knowledge needs to be introduced”). For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” Par. 0064]
wherein the binary classification network model is a model for learning, based on a third sample data set, [Zhao, “the semantic information classification model automatically performs classification predictions by learning (i.e., the claimed “classification network model is a model for learning”) rules from previous tasks (i.e., the claimed “a third sample data set”),” Par. n0096]
a capability of determining whether the external knowledge needs to be introduced, and [Zhao, “The first semantic information vector and the historical dialogue semantic information vector are input into a pre-trained semantic information classification model,” Par. n0014; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term (i.e., the claimed “classification result”), the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term (i.e., the claimed “third classification result”), an information push operation can be performed. The information push operation can call an external knowledge base (i.e., the claimed “whether the external knowledge needs to be introduced”). For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” Par. 0064]
the third sample data set comprises third sample input dialogue data, third sample historical dialogue data, and second sample dialogue environment data corresponding to the third sample input dialogue data. [Zhao, “The first semantic information vector and the historical dialogue semantic information vector are input into a pre-trained semantic information classification model,” Par. n0014; “In addition, fields such as real estate sales, legal consulting, and education consulting often involve a lot of professional knowledge. Therefore, if the slot value is a professional term (i.e., the claimed “classification result”), the corresponding knowledge can be called to push the explanation of the professional term (i.e., the claimed “text knowledge”).” Par. n0079; “If the slot value of the semantic slot in the second semantic information is a location or a professional term (i.e., the claimed “third classification result”), an information push operation can be performed. The information push operation can call an external knowledge base (i.e., the claimed “whether the external knowledge needs to be introduced”). For example, if the slot value is a location, a weather knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push the weather for that location, and a map knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be called to push a map for that location. If the domain is real estate, a real estate knowledge base (i.e., the claimed “external knowledge other than the knowledge graph and the text knowledge”) can be also called to push the housing prices for that location.” Par. n0079; Zhang, “dialogue with the user during multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” Par. 0064]
The combination fails to explicitly teach binary classification.
However, Xu teaches:
wherein the third classification result is obtained by inputting the second historical dialogue data, the current dialogue environment data, and the second input dialogue data into a pre-trained binary classification network model, and [Xu, “The chatbot system can input the utterance into a machine-learning model including a set of binary classifiers (i.e., the claimed pre-trained binary classification network model“).” Par. 0004]
wherein the binary classification network model is a model for learning, based on a third sample data set, [Xu, “The chatbot system can input the utterance into a machine-learning model including a set of binary classifiers (i.e., the claimed pre-trained binary classification network model“).” Par. 0004]
Zhao, Zhang, Feng, Su and Xu pertain to dialogue systems and are analogous to the instant application. Accordingly, it would have been obvious to one of ordinary skill in the dialogue systems art to modify Zhao’s teachings of “dialogue processing method proposed in this application predicts at least the first semantic information corresponding to the current dialogue text based on the current dialogue text; wherein the first semantic information includes: domain information, intent information and semantic slot information; according to a pre-constructed dialogue knowledge graph (i.e., the claimed “second input dialogue data from a user”), the first semantic information and historical dialogue semantic information are fused to obtain second semantic information, wherein the historical dialogue semantic information is semantic information determined based on historical dialogue text (i.e., the claimed “second historical dialogue data”); and according to the second semantic information, dialogue prompt information corresponding to the second semantic information is determined. By employing the technical solution of this application, the first semantic information corresponding to the current dialogue text and the semantic information corresponding to the historical dialogue text are fused together, which can realize dialogue state (i.e., the claimed “current environment data”) tracking, accurately grasp the dependency relationship between each dialogue, and thus determine more accurate useful information, thereby improving the accuracy of dialogue management” (Zhao, Par. n0048) with the teachings of “multiple rounds of question answering (i.e., the claimed “second input dialogue data”)” (Zhang, Par. 0064) taught by Zhang, the teachings of “tactile sense, a hearing sense, a sight sense or a smell sense (i.e., the claimed “current dialogue environment data”) that a person has; a medium of information, including a voice, a video, a text, etc.; various sensors such as a radar, an infrared sensor, an accelerometer, etc.” (Feng, Par. 0078) taught by Feng, the teachings of cross-attention interaction network (i.e., the claimed “cross-attention network model”) and “pre training” (Su,Par. n0005) taught by Su, and the teachings of “binary classifier” (Xu, Par. 0004) taught by Xu in order “create a dialogue experience” and increase “the accuracy as multiple rounds of dialogue proceed” (Zhang, Par. 0064) and to solve the problem of “clarifying a relationship between the knowledge nodes, and an event that actually occurred, and therefore, a problem arises that a semantic understanding result is inaccurate.” (Feng, Par. 0005), “retrieve the best response from several candidate responses” (Su, Par. n0010), and “communicate more intelligently and contextually in live conversations” (Xu, Par. 0003).
Regarding Claim 10, Zhao in view of Zhang, Feng, Su and Xu has been discussed above. The combination further teaches:
wherein the second semantic keyword is obtained by inputting the second historical dialogue data, the current dialogue environment data, and the second input dialogue data into a pre-trained generative network model, and wherein the generative network model is a model for learning, based on a fourth sample data set, a capability of a third sample semantic keyword corresponding to the fourth sample data set, and the fourth sample data set comprises fourth sample input dialogue data, fourth sample historical dialogue data, and third sample dialogue environment data corresponding to the fourth sample input dialogue data. [Claim is directed to repeating the subject matter for another sample data set, another semantic keyword, another input dialogue data and another dialog environment data. However, another sample data set/ another semantic keyword/ another input dialogue data/ another dialog environment data /repeating steps known from prior art with different data is straightforward, amounts to the normal use of the teachings of Zhao in view of Zhang, Feng, Su and Xu and are rejected under similar rationale.]
Regarding Claim 19, Zhao in view of Zhang, Feng, Su and Xu has been discussed above. The combination further teaches:
wherein the first classification result is obtained by inputting the second sample data set into a pre-trained binary classification network model, and wherein the binary classification network model is a model for learning, based on a third sample data set, a capability of determining whether the external knowledge needs to be introduced, and the third sample data set comprises third sample input dialogue data, third sample historical dialogue data, and second sample dialogue environment data corresponding to the third sample input dialogue data. [Xu, “The chatbot system can input the utterance into a machine-learning model including a set of binary classifiers (i.e., the claimed pre-trained binary classification network model“).” Par. 0004; Claim is directed to repeating the subject matter for another sample data set, another semantic keyword, another input dialogue data and another dialog environment data. However, another sample data set/ another semantic keyword/ another input dialogue data/ another dialog environment data /repeating steps known from prior art with different data is straightforward, amounts to the normal use of the teachings of Zhao in view of Zhang, Feng, Su and Xu and are rejected under similar rationale.]
Regarding Claim 20, Zhao in view of Zhang, Feng, Su and Xu has been discussed above. The combination further teaches:
wherein the second sample semantic keyword is obtained by inputting the second sample data set into a pre-trained generative network model, and wherein the generative network model is a model for learning, based on a fourth sample data set, a capability of a third sample semantic keyword corresponding to the fourth sample data set, and the fourth sample data set comprises fourth sample input dialogue data, fourth sample historical dialogue data, and third sample dialogue environment data corresponding to the fourth sample input dialogue data. [Su, “generative dialogue systems (i.e., the claimed “generative network model”), Par. n0002; Claim is directed to repeating the subject matter for another sample data set, another semantic keyword, another input dialogue data and another dialog environment data. However, another sample data set/ another semantic keyword/ another input dialogue data/ another dialog environment data /repeating steps known from prior art with different data is straightforward, amounts to the normal use of the teachings of Zhao in view of Zhang, Feng, Su and Xu and are rejected under similar rationale.]
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Lee et al., (U.S. Patent Application Publication 2023/0134933) teaches knowledge- based dialogue systems.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to EUNICE LEE whose telephone number is 571-272-1886. The examiner can normally be reached M-F 8:00 AM - 5:00 PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Bhavesh Mehta can be reached on 571-272-7453. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/EUNICE LEE/Examiner, Art Unit 2656
/BHAVESH M MEHTA/ Supervisory Patent Examiner, Art Unit 2656
