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 .
Specification
The lengthy specification has not been checked to the extent necessary to determine the presence of all possible minor errors. Applicant’s cooperation is requested in correcting any errors of which applicant may become aware in the specification.
Claim Rejections - 35 USC § 112(b)
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-22 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Regarding claim 1,
Claim 1 recites the limitations updating a user interest record that includes data indicating a level of interest of the user in a set of items, wherein the level of interest for each item is determined. It is unclear whether the phrase “the level of interest for each item” has antecedent basis, as only “a level of interest of the user in a set of items” is earlier recited, with the first phrase having a singular interest level for the collective set of items, and the second phrase having multiple interest levels, one for each item. Thus, the scope of the claim is indefinite. For examination purposes, the limitations will be interpreted as reading “updating a user interest record that includes data indicating a level of interest of the user in a set of items, wherein the level of interest for the items is determined”.
In reference to dependent claims 2-20, claims 2-20 do not cure the deficiencies noted in the rejection of claim 1. In particular claims 2, 8, and 16 also include the phrases of unclear antecedent basis “the level of interest for each item”, “the level of interest of the user in the item”, and “the level of interest for each item”, respectively, which renders the scope of those claims indefinite with equivalent rationale. Therefore, claims 2-20 are rejected under the same rationale as claim 1, and are interpreted for examination purposes in the same way.
Regarding claim 21,
Claim 21 recites a system for performing the function of the method of claim 1. All other limitations in claim 21 are substantially the same as those in claim 1, therefore claim 21 is considered indefinite with an equivalent rationale and is interpreted for examination purposes in the same way.
Regarding claim 22,
Claim 22 recites a medium for performing the function of the method of claim 1. All other limitations in claim 22 are substantially the same as those in claim 1, therefore claim 22 is considered indefinite with an equivalent rationale and is interpreted for examination purposes in the same way.
Prior Art
The following references are used for prior art claim rejections:
Spiegel et al. (U.S. Patent Application Publication No. 2024/0249318), hereinafter Spiegel
Zhou et al. “Deep Interest Network for Click-Through Rate Prediction”, hereinafter Zhou
Schillings et al. (U.S. Patent No. 9,679,062), hereinafter Schillings
Maschmeyer et al. (U.S. Patent Application Publication No. 2024/0256792), hereinafter Maschmeyer
Pi et al. “Practice on Long Sequential User Behavior Modeling for Click-Through Rate Prediction”, hereinafter Pi
Chen et al. “Sequential Recommendation with User Memory Networks”, hereinafter Chen
Yagci et al. “Scalable and adaptive collaborative filtering by mining frequent item co- occurrences in a user feedback stream”, hereinafter Yagci
He and Chu “A Social Network-Based Recommender System (SNRS)”, hereinafter He
Banothu et al. (U.S. Patent Application Publication No. 2020/0045008), hereinafter Banothu
Dai et al. “Uncovering ChatGPT’s Capabilities in Recommender Systems”, hereinafter Dai
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 2, 5, 8, 21, and 22 are rejected under 35 U.S.C. 103 as being unpatentable over Spiegel in view of Zhou.
Regarding claim 1,
Spiegel teaches A method, comprising:
initiating a user session with a conversational user interface of an artificial intelligence system ((Spiegel [0138]) “In operation 402, a user uses an interactive platform application 408, such as application 106 executing on a client system 102 (of FIG. 1), to access the chatbot system 300 so that the user can chat with the chatbot system 300 or add it to existing chats”) that displays, within the conversational user interface, responses to user prompts received during the user session, ((Spiegel [0138]) “The interactive platform application 408 provides the user interface through which the user can interact with the chatbot system 300. Within the interactive platform application 408, the user has the option to initiate a 1-on-1 chat session with the chatbot system 300…From the interactive platform application 408, the user can send textual prompts to the chatbot system 300 and receive conversational responses”)
the responses being generated using one or more machine learning models of the artificial intelligence system; ((Spiegel [0141]]) “In some examples, a more detailed textual response is generated by the chatbot system 300 beyond a short reply. To do so, the LLM 338 generates a longer, more elaborate response”, an LLM is a machine learning model)
during the user session: receiving, by the artificial intelligence system, one or more prompts provided in the conversational user interface by a user; ((Spiegel [0083]) “In operation 302 of chatbot method 374, the chat bot system 300 receives, from a client system 336, a prompt 328 of a user during a first interactive session”)
displaying, in the conversational user interface, one or more digital components ((Spiegel [0209]) “In some examples, implementation of the chatbot system 300 in a conversational interface integrated within an interactive application provides messaging, content, and display advertisements (including AI-generated ads based on keyword winning links) which also helps provide signals for ranking of all content provided by the chatbot system 300 (including display advertisements, media, and AR)”, the specification states at [0034] “an advertisement is a type of digital component”) that each include content related to a corresponding item based at least in part on the one or more prompts; (Spiegel Fig. 3B shows content related to a hotel item in response to a user prompt, within a conversational user interface, at 336)
PNG
media_image1.png
797
1070
media_image1.png
Greyscale
detecting, for each displayed digital component, one or more user interaction events indicative of whether the user interacted with the displayed digital component while the digital component was displayed in the conversational user interface; ((Spiegel [0121]) “In operation 326, the chatbot system 300 modifies delivery of advertising content based on user engagement with the advertising content. For example, the chatbot system 300 modifies the delivery of advertising content based on user engagement with the ads. Specifically, the chatbot system 300 tracks metrics such as click-through rates, conversion rates, and dwell time for the generated ads”)
Zhou teaches the following further limitations that Spiegel does not explicitly teach:
updating a user interest record that includes data indicating a level of interest of the user in a set of items, ((Zhou Pg. 2) “we propose a novel model: Deep Interest Network (DIN), which adaptively calculates the representation vector of user interests by taking into consideration the relevance of historical behaviors given a candidate ad”, (Zhou Pg. 3) “In e-commerce sites, such as Alibaba, advertisements are natural goods. In the rest of this paper, without special declaration, we regard ads as goods”)
wherein the level of interest for each item is determined by the one or more machine learning models based on the one or more user interaction events for each digital component; ((Zhou Pg. 5) “Imagine when the young mother mentioned above in section 3 visits the e-commerce site, she finds the displayed new handbag cute and clicks it. Let’s dissect the driving force of click action. The displayed ad hits the related interests of this young mother by soft-searching her historical behaviors and finding that she had browsed similar goods of tote bag and leather handbag recently. In other words, behaviors related to displayed ad greatly contribute to the click action”)
and displaying one or more additional digital components in the [conversational] user interface based at least in part on the user interest record ((Zhou Pg. 3) “Figure 1 briefly illustrates the running procedure of display advertising system in Alibaba, which consists of two main stages: i) matching stage which generates list of candidate ads relevant to the visiting user via methods like collaborative filtering, ii) ranking stage which predicts CTR for each given ad and then selects top ranked ones”, Zhou Pg. 3, Fig. 1 shows advertisements being displayed in a user interface, Spiegel but not Zhou teaches a specifically conversational user interface)
PNG
media_image2.png
323
577
media_image2.png
Greyscale
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel and Zhou by taking the method for displaying digital components including content related to corresponding items within a conversational user interface, where user interaction with the digital components is tracked, taught by Spiegel, and updating a user interest record that tracks user interest in a set of items, then displaying additional digital components in a user interface based on the updated user interest record, taught by Zhou, as doing so would increase user engagement with the digital components by having them match the user’s interests more closely, with the predictable effect of increasing the effectiveness of the digital components. Such a combination would be obvious.
Regarding claim 2,
Spiegel and Zhou jointly teach The method of claim 1,
Spiegel further teaches:
wherein the level of interest for each item determined by the one or more machine learning models is further based on information ((Spiegel [0111]) “the chatbot system 300 furthers a conversation with a user by knowing that user intent (e.g., if the user asks for good hotels in Cancun, the chatbot system 300 responds with: ‘Here is a list of hotels. I also know of a good promotion for a hotel, would you like to see it?’). This provides for the user intent being extracted from the user, matching the intent to possible ads, and embedding that knowledge into a response”, a hotel is an item and intent corresponds to interest) related to one or more previous user sessions of the user with the artificial intelligence system ((Spiegel [0035]) “In some examples, a chatbot system receives a prompt from a user during a first interactive session. The chatbot system generates a response using the prompt and a large language model. The chatbot system communicates the response to the user during the first interactive session. The chatbot system determines a user intent based on the user prompt and response. The chatbot system determines advertising content based on the user intent. The chatbot system communicates the advertising content to the user during a second interactive session”, a first interactive session is a previous user session)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Spiegel and Zhou for the parent claim of claim 2, claim 1. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Regarding claim 5,
Spiegel and Zhou jointly teach The method of claim 1,
Spiegel further teaches:
wherein the user interest record is maintained across a plurality of user sessions, ((Spiegel Abstract) “The extracted user intent, comprising weighted keywords and concepts, is used to create a user intent profile. Targeted advertising content is generated based on the user intent profile and provided to the user during subsequent platform interactions”, a user intent profile corresponds to a user interest record, using a previously generated user intent profile during subsequent platform interactions corresponds to maintaining the user intent profile across a plurality of sessions)
and wherein the user interest record is updated during each user session of the plurality of user sessions ((Spiegel [0092]) “the chatbot system 300 stores a conversation state in the user profile database 354 of a series of interactive sessions so that the chatbot system 300 can have a context for conversations that occur over a plurality of interactive sessions”, (Spiegel [0156]-[0157]) “Operations of the dynamic interactive platform ads module 424 comprise:…Analyzing the conversation context and user profile to determine user interests and intent”, a conversation state stored in a user profile database over a series of sessions corresponds to updating a user interest record during the plurality of user sessions)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Spiegel and Zhou for the parent claim of claim 5, claim 1. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Regarding claim 8,
Spiegel and Zhou jointly teach The method of claim 1,
Zhou further teaches:
wherein data indicating a level of interest of the user in a set of items comprises a weight for each item in the set of items, ((Zhou Pg. 5) “DIN introduces a novel designed local activation unit and maintains the other structures the same. Specifically, activation units are applied on the user behavior features, which performs as a weighted sum pooling to adaptively calculate user representation given a candidate ad A, as shown in Eq. (3)”)
PNG
media_image3.png
47
403
media_image3.png
Greyscale
wherein the weight for each item is indicative of the level of interest of the user in the item ((Zhou Pg. 5) “value of w is treated as an approximation of the intensity of activated user interests to some degree. For example, if one user’s historical behaviors contain 90% clothes and 10% electronics. Given two candidate ads of T-shirt and phone, T-shirt activates most of the historical behaviors belonging to clothes and may get larger value of (higher intensity of interest) than phone”)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Spiegel and Zhou for the parent claim of claim 5, claim 1. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Regarding claim 21,
Claim 21 recites a system comprising one or more processors and one or more storage devices with instructions for performing the function of the method of claim 1. Specifically, claim 21 recites one or more processors; and one or more storage devices storing instructions that, when executed by the one or more processors, cause the one or more processors to carry out operations comprising: [the method of claim 1]. Spiegel recites: (Spiegel [0324]) “The machine 1100 may include processors 1104, memory 1106…the processors 1104…may include, for example, a processor 1112 and a processor 1114 that execute the instructions 1102. The term ‘processor’ is intended to include multi-core processors that may comprise two or more independent processors”.
All other limitations in claim 21 are substantially the same as those in claim 1, therefore the same rationale for rejection applies.
Regarding claim 22,
Claim 22 recites a non-transitory computer readable storage medium with instructions for performing the function of the method of claim 1. Specifically, claim 22 recites A non-transitory computer readable storage medium carrying instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising: [the method of claim 1]. Spiegel recites: (Spiegel [0355]) “Example 8 is at least one machine-readable medium including instructions that, when executed by processing circuitry, cause the processing circuitry to perform operations to implement any of Examples 1-7”.
All other limitations in claim 22 are substantially the same as those in claim 1, therefore the same rationale for rejection applies.
Claims 3, 7, and 9-12 are rejected under 35 U.S.C. 103 as being unpatentable over Spiegel, in view of Zhou, further in view of Schillings.
Regarding claim 3,
Spiegel and Zhou jointly teach The method of claim 1, wherein updating the user interest record comprises:
Schillings teaches the following further limitations that Spiegel does not teach and more explicitly than Zhou teaches:
including the data of the user interest record in one or more prompts ((Schillings Col. 9, lines 61-64) “the initial user profile 112 is employed by the user device 110 for generating recommendations during the early use of the recommender module 150”, a prompt corresponds to input to a machine learning model) to the one or more machine learning models; ((Schillings Col. 7, lines 37-42) “the recommendation module 150 can be configured with various artificial intelligence features so that with continued usage, it can learn more about the attributes and habits of the user and update the initial user profile 112 to stay current with changing user habits and preferences”, artificial intelligence features to enable learning from changing data correspond to use of a machine learning model)
and receiving a recommended update to the user interest record from the one or more machine learning models ((Schillings Col. 7, lines 37-42) “the recommendation module 150 can be configured with various artificial intelligence features so that with continued usage, it can learn more about the attributes and habits of the user and update the initial user profile 112 to stay current with changing user habits and preferences”)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, and Schillings by taking the method of claim 1 for updating a user interest record for a set of items, in order to subsequently display digital components, taught by Spiegel and Zhou, and additionally providing the user interest record to a machine learning model to subsequently receive an update to the user interest record, taught by Schillings, as Schillings teaches that doing so enables (Schillings Col. 7, lines 40-44) “user profile 112 to stay current with changing user habits and preferences. This can mitigate the recommendation module 150 providing recommendations from stale user profile information”, providing the predictable effect of maintaining a continuously accurate user interest record. Such a combination would be obvious.
Regarding claim 7,
Spiegel and Zhou jointly teach The method of claim 1, further comprising, after displaying the one or more additional digital components in the conversational user interface:
Schillings teaches the following further limitations more explicitly than Spiegel or Zhou teaches:
detecting, for each additional displayed digital component, one or more user interaction events indicative of whether the user interacted with the displayed additional digital component while the additional digital component was displayed in the conversational user interface; ((Schillings Col. 12, lines 28-35) “The update module 410, also comprised in the recommendation module 150 on the user device 110, is configured to determine if there was any user interaction with the content item included in the contextual display generated by the presentation module 408. For example, user selection of the content item for further review or user input hiding the content item or even no user interaction can be recorded by the update module 410 for applying to the user profile 152”)
and updating the user interest record based on the one or more user interaction events for each additional digital component ((Schillings Col. 12, lines 36-38) “Based on the user interactions the update module 410 can be further configured to update the user profile stored on the local data store 160 of the user device 110”)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, and Schillings by taking the method of claim 1 for updating a user interest record for a set of items, in order to subsequently display digital components, taught by Spiegel and Zhou, and additionally detecting user interaction with the digital component and subsequently further updating the user interest record, taught by Schillings, as doing so would increase the accuracy of the user interest record by updating it based on empirically measured interaction with digital components reflecting measured user interest, allowing the user interest record to be used to select digital components that a user is more likely to engage with, with the predictable effect of increasing the effectiveness of the digital components. Such a combination would be obvious.
Regarding claim 9,
Spiegel and Zhou jointly teach The method of claim 8, wherein updating the user interest record comprises:
Schillings teaches the following further limitations that Spiegel does not teach and more explicitly than Zhou teaches:
including the data of the user interest record in one or more prompts ((Schillings Col. 9, lines 61-64) “the initial user profile 112 is employed by the user device 110 for generating recommendations during the early use of the recommender module 150”, a prompt corresponds to input to a machine learning model) to the one or more machine learning models; ((Schillings Col. 7, lines 37-42) “the recommendation module 150 can be configured with various artificial intelligence features so that with continued usage, it can learn more about the attributes and habits of the user and update the initial user profile 112 to stay current with changing user habits and preferences”, artificial intelligence features to enable learning from changing data correspond to use of a machine learning model)
and receiving a recommended update to the weights of the user interest record from the one or more machine learning models ((Schillings Col. 14, lines 25-29) “FIG. 8 is a flowchart 800 detailing a method of updating a user profile in accordance with one embodiment. The method begins at 802 with the recommendation module 150 detecting user interaction with the content item presented in the contextual display”, (Schillings Col. 14, lines 39-48) “if the user interaction is indicative of positive user interest the topics, categories and keywords associated with the content item are obtained and if already included in the user profile, their weights are increased. If any of the topics, categories or keywords are not included in the user profile they may be added to the user profile. Similarly, if the user expresses indifference or dislikes the content item, the weights associated with the topics or keywords of the content item in the user profile are decreased”, the recommendation module updates user profiles, and the recommendation module is stated to use artificial intelligence, which corresponds to machine learning)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, and Schillings by taking the method of claim 8 for updating a user interest record for a set of items, comprising weights for items, in order to subsequently display digital components, taught by Spiegel and Zhou, and additionally detecting user interaction with the digital component and subsequently further updating the weights of the user interest record, taught by Schillings, as doing so would increase the accuracy of the user interest record by updating it based on empirically measured interaction with digital components reflecting measured user interest, allowing the user interest record to be used to select digital components that a user is more likely to engage with, with the predictable effect of increasing the effectiveness of the digital components. Such a combination would be obvious.
Regarding claim 10,
Spiegel and Zhou jointly teach The method of claim 8,
Schillings teaches the following further limitations that Spiegel does not teach and more explicitly than Zhou teaches:
wherein updating the user interest record comprises updating, for each user interaction event with each displayed digital component, the weight for the item for which the digital component includes content (Schillings Col. 14, lines 32-48) “At 804 the type of user interaction is determined to indicate user interest or disinterest in the content item. For example, it is determined at 804 if the user interaction is indicative of positive user interest or the user actively dislikes the content or if the user is indifferent to the content item. Based on the type of user interest, the user profile locally stored on the user device 110 is updated at 806. In one embodiment, if the user interaction is indicative of positive user interest the topics, categories and keywords associated with the content item are obtained and if already included in the user profile, their weights are increased. If any of the topics, categories or keywords are not included in the user profile they may be added to the user profile. Similarly, if the user expresses indifference or dislikes the content item, the weights associated with the topics or keywords of the content item in the user profile are decreased”)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, and Schillings by taking the method of claim 8 for updating a user interest record for a set of items, comprising weights for items, in order to subsequently display digital components, taught by Spiegel and Zhou, and additionally detecting user interaction with the digital component and subsequently further updating an item weight of the user interest record corresponding to the digital component, taught by Schillings, as doing so would increase the accuracy of the user interest record by updating it based on empirically measured interaction with digital components reflecting measured user interest, allowing the user interest record to be used to select digital components that a user is more likely to engage with, with the predictable effect of increasing the effectiveness of the digital components. Such a combination would be obvious.
Regarding claim 11,
Spiegel, Zhou, and Schillings jointly teach The method of claim 10,
Schillings further teaches:
wherein updating the weight for the item for which the digital component includes content comprises updating the weight by a different amount based on a type of the one or more user interaction events with the displayed digital component (Schillings Col. 14, lines 39-48) “In one embodiment, if the user interaction is indicative of positive user interest the topics, categories and keywords associated with the content item are obtained and if already included in the user profile, their weights are increased. If any of the topics, categories or keywords are not included in the user profile they may be added to the user profile. Similarly, if the user expresses indifference or dislikes the content item, the weights associated with the topics or keywords of the content item in the user profile are decreased”, interaction indicating positive user interest and interaction indicating negative user interest are types of user interaction)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Spiegel, Zhou, and Schillings for the parent claim of claim 11, claim 10. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Regarding claim 12,
Spiegel and Zhou jointly teach The method of claim 8,
Schillings teaches the following further limitations that Spiegel does not teach and that Zhou does not teach explicitly:
wherein updating the user interest record comprises reducing, for each displayed digital component for which the user interaction event indicates that the user did not interact with the digital component, the weight for the item for which the digital component includes content ((Schillings Col. 14, lines 32-39) “At 804 the type of user interaction is determined to indicate user interest or disinterest in the content item. For example, it is determined at 804 if the user interaction is indicative of positive user interest or the user actively dislikes the content or if the user is indifferent to the content item. Based on the type of user interest, the user profile locally stored on the user device 110 is updated at 806”, (Schillings Col. 14, lines 45-48) “Similarly, if the user expresses indifference or dislikes the content item, the weights associated with the topics or keywords of the content item in the user profile are decreased”)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, and Schillings by taking the method of claim 8 for updating a user interest record for a set of items, comprising weights for items, in order to subsequently display digital components, taught by Spiegel and Zhou, and additionally detecting a lack of interaction with digital components and reducing the weights of items corresponding to the digital components, taught by Schillings, as doing so would increase the accuracy of the user interest record by updating it based on empirically measured interaction with digital components reflecting measured user interest, allowing the user interest record to be used to select digital components that a user is more likely to engage with, with the predictable effect of increasing the effectiveness of the digital components. Such a combination would be obvious.
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over Spiegel, in view of Zhou, further in view of Maschmeyer.
Regarding claim 4,
Spiegel and Zhou jointly teach The method of claim 1,
Maschmeyer teaches the following further limitations more explicitly than Spiegel or Zhou teaches:
wherein displaying one or more additional digital components in the [conversational] user interface comprises providing the data of the user interest record with a request for a digital component to the artificial intelligence system ((Maschmeyer [0021]) “responsive to a request for a textual description, retrieving a user record for a user associated with the request; obtaining one or more user attributes based on the user record; generating a prompt to a large language model (LLM) for generating a user-specific textual description, the prompt including the one or more user attributes to include in the generated user-specific textual description and a source text; providing the prompt to the LLM and receive a generated user-specific textual description; and providing the generated user-specific textual description for display via a user device”, Spiegel teaches a specifically conversational user interface)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, and Maschmeyer by taking the method of claim 1 for updating a user interest record for a set of items, in order to subsequently display digital components, taught by Spiegel and Zhou, and additionally providing the user interest record to an artificial intelligence system with a request for a digital component, taught by Maschmeyer, as Maschmeyer teaches: (Maschmeyer [0085]) “the prompt 535 generated by the prompt generator 500 includes instructions to the LLM to include one or more user attributes 525 in the generated textual description so that the generated textual description is a personalized textual description 575 that is customized to the user” and (Maschmeyer [0099]) “user attributes 525 may be extracted from user record(s) 505”. Such a combination would be obvious.
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Spiegel, in view of Zhou, further in view of Pi.
Regarding claim 6,
Spiegel and Zhou jointly teach The method of claim 1,
Pi teaches the following further limitations that neither Spiegel nor Zhou teaches:
further comprising constraining a data size of the user interest record ((Pi Pg. 6) “Both NTM and MIU module enjoy the benefit of incremental calculation. The latest memory states represent user interests and are updated to TAIR for realtime CTR prediction. When receiving a new user behavior event, UIC calculates and updates the user interest representation to TAIR again. In this way, user behavior data is not needed to be stored. The huge volume of longtime user behaviors could be reduced from 6T to 2.7T in our system”)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, and Schillings by taking the method of claim 1 for updating a user interest record for a set of items, in order to subsequently display digital components, taught by Spiegel and Zhou, and additionally constraining a data size of the user interest record, taught by Pi, as Pi teaches (Pi Pg. 3) “high performance storage is used in our system to keep low latency and high throughput, and it is too expensive to hold such huge storage. The huge volume also causes corresponding calculation and update of user behavior features to be costly enough. Thus a fairly long behavior sequence means a unacceptable storage consumption”, that is, that a user interest record that is too large results in high latency and high storage costs, both of which are undesirable, and constraining the size of the user interest record provides the predictable benefit of mitigating these. Such a combination would be obvious.
Claims 13 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Spiegel, in view of Zhou, further in view of Chen.
Regarding claim 13,
Spiegel and Zhou jointly teach The method of claim 1,
Chen teaches the following further limitations that neither Spiegel, nor Zhou teaches:
wherein updating the user interest record comprises adding one or more additional items to the set of items based on the one or more user interaction events ((Chen Pg. 4) “As mentioned above, users’ recent behaviors usually play more important roles for current predictions. As a result, we adopt a simple first-in-first-out mechanism to maintain the latest interacted items in the user memory matrix Mu. Specifically, the memory matrix Mu for user u always stores the most recent K items…When writing the memory, the earliest item would be replaced, and Mu is updated…Note that when the memory is not full, the item is directly added without replacing any other entry”)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, and Chen by taking the method of claim 1 for updating a user interest record for a set of items based on user interaction, in order to subsequently display digital components, taught by Spiegel and Zhou, and additionally adding items to the set of items based on the user interaction, taught by Chen, as Chen teaches (Chen Pg. 1) “As exampled in Figure 1(a), the key reason for a user to buy a phone case (item E) could be that he bought a phone (item B) before” and (Chen Pg. 5) “users’ recent behaviors usually play more important roles for current predictions”, that is, that users’ interests and corresponding items can change over time, and adding items to a set of items in a user interest record provides the predictable benefit of increasing the flexibility of representing these changing interests. Such a combination would be obvious.
Regarding claim 14,
Spiegel and Zhou jointly teach The method of claim 1,
Chen teaches the following further limitations that neither Spiegel, nor Zhou teaches:
wherein updating the user interest record comprises removing an item from the set of items based on the user interaction event [for a displayed digital component] indicating that the user did not interact [with the displayed digital component] ((Chen Pg. 4) “As mentioned above, users’ recent behaviors usually play more important roles for current predictions. As a result, we adopt a simple first-in-first-out mechanism to maintain the latest interacted items in the user memory matrix Mu. Specifically, the memory matrix Mu for user u always stores the most recent K items…When writing the memory, the earliest item would be replaced, and Mu is updated”, an item not being interacted with for a period of time corresponds to an indication that the user did not interact with the item, displayed digital components for items are taught by Spiegel and Zhou)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, and Chen by taking the method of claim 1 for updating a user interest record for a set of items based on user interaction, in order to subsequently display digital components, taught by Spiegel and Zhou, and additionally removing items from the set of items based on lack of user interaction, taught by Chen, as Chen teaches (Chen Pg. 5) “users’ recent behaviors usually play more important roles for current predictions”, that is, that users’ interests and corresponding items can change over time, and removing items previously present in a user interest record based on lack of continued interact provides the predictable benefit of increasing the flexibility of representing these changing interests. Such a combination would be obvious.
Claim 15 is rejected under 35 U.S.C. 103 as being unpatentable over Spiegel, in view of Zhou, further in view of Banothu.
Regarding claim 15,
Spiegel and Zhou jointly teach The method of claim 1,
Banothu teaches the following further limitations that neither Spiegel, nor Zhou teaches:
further comprising, during the user session, updating ((Banothu Abstract) “Content items provided by the online system include links to an online communication system from which the online system can directly observe whether the user engaged in communications with third parties. The online system can thus obtain labeled training data describing communications between users and users' previous interactions with content items and pages of the online system”) a positive user interest record based on the one or more user interaction events for a displayed digital component indicating that the user interacted with the displayed digital component, ((Banothu [0045]) “In an embodiment, the model training module 240 forms a training set of content items by identifying a positive set of content items for which a user engaged with the content item (e.g., by clicking on it) and subsequently sent at least one message in a chat conversation with the content provider. The message may be any message sent to the chat conversation or may be a message having specified characteristics, as discussed above”)
and updating a negative user interest record based on the user interaction event for a displayed digital component indicating that the user did not interact with the displayed digital component ((Banothu [0045]) “In some embodiments, the model training module 240 also trains a negative set of content items for which the user did not engage with the content item or did not send at least one message in a chat conversation with the content provider”)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, and Banothu by taking the method of claim 1 for detecting user interaction with displayed digital components, taught by Spiegel and Zhou, and additionally updating a positive user interest record when a user interacts with a digital component and a negative user interest record when a user does not interact with a digital component, taught by Banothu, as Banothu teaches (Banothu [0004]) “Users often ignore content items sent by the online system, and as a result, the online system wastes resources by sending the ignored content items. Content providers would prefer to send content to users that are likely to interact with the content” and collecting positive and negative user interest records allows for a model to be trained with positive and negative examples, so that subsequently ((Banothu Abstract) “The trained model is applied to candidate content items to predict a probability that a user will engage in a communication with the content provider if the content is provided to the user, and the model optimizes the selection of content items for which the call to action is to engage in the communication”). Such a combination would be obvious.
Claims 16 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Spiegel, in view of Zhou, further in view of Yagci.
Regarding claim 16,
Spiegel and Zhou jointly teach The method of claim 1, wherein the method further comprises:
Spiegel further teaches:
initiating a second user session with the conversational user interface; ((Spiegel [0203]) “In some examples, a user uses "@" to add the chatbot system 300 to a Chat and directly summons the 300 to help with a query”, adding a chatbot to an existing chat corresponds to initiating a second user session with a conversational user interface)
during the second user session: receiving, by the artificial intelligence system, one or more second prompts provided in the conversational user interface by a second user; ((Spiegel [0202]-[0203]) “In some examples, the user may interact with the chatbot system 300 directly by exchanging user prompts and chatbot responses directly. In some examples, two or more users may interact with one another while the chat bot system 300 monitors the interaction in order to discern user intent of one or more of the users…A user taps the chatbot icon 504 on the presence bar to start asking the chatbot system 300 a question or writes @ChatBot in the chat bar”, a question from a user in a group chat comprising multiple users corresponds to a second prompt from a second user)
displaying, in the conversational user interface, one or more second digital components ((Spiegel [0156]) “In some examples, the skill modules 422 comprise dynamic interactive platform ads module 424. The dynamic interactive platform ads module 424 provides relevant, targeted advertisements to users during a conversation”) that each include content related to a corresponding item based at least in part on the one or more second prompts; (Spiegel Fig. 3B shows content related to a hotel item in response to a user prompt, within a conversational user interface, at 336)
detecting, for each displayed second digital component, one or more second user interaction events indicative of whether the second user interacted with the second displayed digital component while the second digital component was displayed in the conversational user interface; ((Spiegel [0121]) “In operation 326, the chatbot system 300 modifies delivery of advertising content based on user engagement with the advertising content. For example, the chatbot system 300 modifies the delivery of advertising content based on user engagement with the ads. Specifically, the chatbot system 300 tracks metrics such as click-through rates, conversion rates, and dwell time for the generated ads”)
Zhou further teaches:
and wherein the [second] set of items includes the corresponding items of each [second] digital component; ((Zhou Pg. 3) “In e-commerce sites, such as Alibaba, advertisements are natural goods. In the rest of this paper, without special declaration, we regard ads as goods. Figure 1 briefly illustrates the running procedure of display advertising system in Alibaba, which consists of two main stages: i) matching stage which generates list of candidate ads relevant to the visiting user via methods like collaborative filtering, ii) ranking stage which predicts CTR for each given ad and then selects top ranked ones”, Zhou does not explicitly teach second digital components or a second set of items for a second user)
Yagci teaches the following further limitations that neither Spiegel, nor Zhou explicitly teaches:
updating an additional user interest record that includes data indicating a level of interest of the second user in a second set of items, ((Yagci Pgs. 4-5) “a list of size |U| is maintained which holds a separate set of items for each user's interaction history. We call this structure the user list, UL, where ULu denotes the set of items, user u has already interacted with…By making use of UL and IL, the stream is processed as follows: With the arrival of each 〈u, i, t〉, stream summary of each item previously assessed by user u is updated by inserting item i”)
wherein the level of interest for each item is determined by the one or more machine learning models based on the one or more second user interaction events for each second digital component, ((Yagci Pg. 4) “Let S define a stream of tuples 〈u, i, t〉, where u ∈ U, i ∈ I, and t is a timestamp. We assume that a tuple represents a positive assessment of item i by user u, such as a like, a click, a purchase, or a binarized rating”, Zhou teaches that items themselves can be digital components, clicking an item corresponds to a user interaction event for the corresponding digital component)
and updating the user interest record based on the additional user interest record ((Yagci Pg. 6) “personalized recommendation to a queried user, u, can be performed with the up-to-date UL and IL…For each item i ∈ ULu, its top-k frequently co-occurring items are returned from ILi. If a returned item is not already in ULu, then it is assumed to be a candidate item, ic, for recommendation…In the end, each candidate item, ic, is assumed to have a total similarity to a user's current assessments…top-N recommendations are returned by sorting candidate items in descending order with respect to their total similarities”, assigning recommendations to a user based on an updated interest record of another user corresponding to updating a user interest record based on an additional user interest record)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, and Yagci by taking the method of claim 1 for updating a user interest record for a set of items based on detected user interaction, each item in the set of items corresponding to a digital component, in order to subsequently display additional digital components, and additionally repeating the detection of user interaction for a second user, taught by Spiegel and Zhou, and including updating a second user interest record in a set of items for a second user based on interest shown in the items corresponding to digital components, and updating the first user’s user record based on the second user record, taught by Yagci, as Yagci teaches that their method (Yagci Pg. Abstract) “offers adaptivity in the sense that newly arriving user-item interactions are immediately integrated into the model. The model is always up-to-date and it can readily be used to recommend items to users with the most recent information” and that (Yagci Pg. 1) “Collaborative filtering can be used for personalized or item-to-item recommendations. Ways for achieving personalization include recommending to a user top-N similar items to her item preference history, or recommending to a user items by looking at item preferences of similar users”, and thus updating one user’s interests based on an update to another user’s interest allows for maximally effective recommendation of items with the most recent data collected. Such a combination would be obvious.
Regarding claim 17,
Spiegel, Zhou, and Yagci jointly teach The method of claim 16,
Yagci further teaches:
wherein updating the user interest record comprises adding one or more items from the second set of items to the set of items ((Yagci Pg. 4) “a list of size |U| is maintained which holds a separate set of items for each user's interaction history. We call this structure the user list, UL, where ULu denotes the set of items, user u has already interacted with”, (Yagci Pg. 6) “personalized recommendation to a queried user, u, can be performed with the up-to-date UL and IL…For each item i ∈ ULu, its top-k frequently co-occurring items are returned from ILi. If a returned item is not already in ULu, then it is assumed to be a candidate item, ic, for recommendation…In the end, each candidate item, ic, is assumed to have a total similarity to a user's current assessments…top-N recommendations are returned by sorting candidate items in descending order with respect to their total similarities”, assigning recommendations to a user from a set of items of all users, which is updated whenever another user interacts with an item, corresponds to adding at least one item from a second user’s set of items to the user’s set of items)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Spiegel, Zhou, and Yagci for the parent claim of claim 17, claim 16. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Claims 18 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Spiegel, in view of Zhou, further in view of Yagci, further in view of He.
Regarding claim 18,
Spiegel, Zhou, and Yagci jointly teach The method of claim 16,
He teaches the following further limitations that Spiegel does not teach explicitly and that Zhou and Yagci do not teach:
wherein the second user is connected to the user ((He Pg. 9) “SNRS learns the correlations between the target user U and each of his/her immediate friends V from the items that they both have rated previously, and then assume each pair of friends will behave consistently on reviewing the target item I too. Thus, U’s rating can be estimated from rVI according to the correlations”) in a social network ((He Abstract) “We extract data from a real online social network, and our analysis of this large data set reveals that friends have a tendency to select the same items and give similar ratings”)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, Yagci, and He by taking the method of claim 16, where one user’s interest record is updated based on an update to a second user’s interest record, taught by Spiegel, Zhou, and Yagci, and including the first and second user being connected within a social network, taught by He, as He teaches (He Pg. 1) “We extract data from a real online social network, and our analysis of this large data set reveals that friends have a tendency to select the same items and give similar ratings. Experimental results on this data set show that our proposed system not only improves the prediction accuracy of recommender systems but also remedies the data sparsity and cold-start issues inherent in collaborative filtering”. Such a combination would be obvious.
Regarding claim 20,
Spiegel, Zhou, and Yagci jointly teach The method of claim 16,
He teaches the following further limitations that neither Spiegel, nor Zhou, nor Yagci teaches:
wherein updating the user interest record further comprises determining that the one or more second user interaction events are similar to the one or more user interaction events ((He Pg. 4) “CF will then make recommendations to the target user by predicting the target user’s rating to the target item based on the ratings of his/her top-K similar users”, similar ratings are similar interaction events)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, Yagci, and He by taking the method of claim 16, where one user’s interest record is updated based on an update to a second user’s interest record, taught by Spiegel, Zhou, and Yagci, and including the first and second user having similar interaction events, taught by He, as a user’s interest in a digital component can be predicted based on a similar user’s interest in a similar digital component, providing the predictable benefit of increasing the likelihood of selecting a digital component that the first user will engage with. Such a combination would be obvious.
Claim 19 is rejected under 35 U.S.C. 103 as being unpatentable over Spiegel, in view of Zhou, further in view of Yagci, further in view of Dai.
Regarding claim 19,
Spiegel, Zhou, and Yagci jointly teach The method of claim 6,
Dai teaches the following further limitations that neither Spiegel, nor Zhou, nor Yagci teaches:
wherein updating the user interest record ((Dai Pg. 4) “we employ prompt tuning to elicit three ranking capabilities from LLMs. As depicted in Figure 1 (left), the prompt of In-Context Learning consists of three components:…(iii) New input query of a given user f(h’, c’ | u), which needs to be answered by LLMs…As depicted in Figure 1 (right), LLMs will utilize the different ranking capabilities elicited through different prompts to make predictions y-hat’”) further comprises determining that the one or more second prompts are similar to the one or more prompts ((Dai Pg. 4) “we employ prompt tuning to elicit three ranking capabilities from LLMs. As depicted in Figure 1 (left), the prompt of In-Context Learning consists of three components:…(ii) Demonstration examples D ={f(h1, c1, y1), …, f(hN, cN, yN)} (i.e., N-shot in-context learning), where h denotes the historical interacted items of a user, c denotes the candidate items which need to be ranked, y denotes the predictions given by LLMs, and f(·) is the function for transforming the examples into designed prompt templates. The demonstration examples facilitate the LLM’s comprehension of the current task”, ranking new items for a prompt for a new user based on similarity to an existing prompt corresponds to updating a user interest record based on determination of similarity between first and second prompts)
At the time of filing, one of ordinary skill in the art would have motivation to combine Spiegel, Zhou, Yagci, and Dai by taking the method of claim 16, where one user’s interest record is updated based on an update to a second user’s interest record, taught by Spiegel, Zhou, and Yagci, and including the first and second user having similar prompts, taught by Dai, as a user’s interest in an item that can correspond to a digital component given a prompt from the user can be predicted based on a similar user’s prompt, providing the predictable benefit of increasing the likelihood of selecting a digital component that the first user will engage with. Such a combination would be obvious.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Dorai-Raj et al. (U.S. Patent Application Publication No. 2020/0226418) teaches a method for updating a UI to include a UI element with a high user interest score determined from interaction data.
Shalom et al. “Beyond Collaborative Filtering: The List Recommendation Problem” teaches a method for recommending lists to users to maximize click probability.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to VICTOR A NAULT whose telephone number is (703) 756-5745. The examiner can normally be reached M - F, 12 - 8.
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, Miranda Huang can be reached at (571) 270-7092. 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.
/V.A.N./Examiner, Art Unit 2124
/MIRANDA M HUANG/Supervisory Patent Examiner, Art Unit 2124
Prosecution Insights