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 .
DETAILED ACTION
The Action is responsive to the Amendments and Remarks filed on 4/16/2026. Claims 1-20 are pending claims. Claims 1, 8, and 15 are written in independent form.
Claim Rejections - 35 USC § 112
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 1-20 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 Claims 1, 8, and 15 recite the limitations “creating a semantic cache…based on training data with a plurality of training samples each of which includes a past query and representing an intent associated with the past query” and “wherein the semantic cache includes a plurality of triplets, each of which includes a past query, an intent related to the past query, and a topic associated with the past query” renders the claims indefinite because it is unclear if the “a past query” included in the training samples to create the semantic cache is the same as the “a past query” and “the past query” recited with respect to the plurality of triplets included in the semantic cache.For the purpose of prosecution herein, the claims are being understood where “an intent associated with the past query” is referring to “a past query” included in the training samples and “an intent related to the past query, and a topic associated with the past query” is referring to “a past query” included in the semantic cache, different from the “a past query” in the training samples. It is further understood that the “past query” recited as being included in the training samples can be the same or different from the “past query” recited as being included in the semantic cache.
Dependent Claims 2-7, 9-14, and 16-20 inherit the deficiencies of their parent claims and are therefore being rejected based upon the same reason(s) stated for their parent claims.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-patentable subject matter. The claimed invention is directed to one or more abstract ideas without significantly more. The judicial exception is not integrated into a practical application. The claims do not include additional elements that are sufficient to amount to significantly more than judicial exception. The eligibility analysis in support of these findings is provided below.
As per Claims 1, 8, and 15,
STEP 1:In accordance with Step 1 of the eligibility inquiry (as explained in MPEP 2106), the claimed method (claims 1-7), machine readable and non-transitory medium (claims 8-14), and system (claims 15-20) are directed to one of the eligible categories of subject matter and therefore satisfies Step 1.
STEP 2A Prong One:The independent claims 1, 8, and 15 recite the following limitations directed to an abstract idea:
Estimating an intent and a topic associated with the query based on the semantic cache; and
The limitation recites a mental process of observation, evaluation, judgement, and/or opinion capable of being performed by the human mind, or by a human using a pen and paper, by observing and evaluating the query and the semantic cache, and based on the observation and evaluation, making a judgement and/or opinion of an estimated intent and topic associated with the query.
Providing the estimated intent and the estimated topic to the query-based application to facilitate execution of the query-based task in accordance with both the query and the estimated intent and the topic of the query associated therewith.
The limitation recites a mathematical concept of executing a mathematical formula/function in the form of inserting the estimated intent and the estimated topic as input “to the query-based application to facilitate execution of the query-based task in accordance with both the query and the estimated intent and the topic of the query associated therewith”.
STEP 2A Prong Two:Claim 8 recites that the steps are performed using “a machine-readable and non-transitory medium” and “a machine”, which is a high-level recitation of generic computer components and represents mere instructions to apply on a computer as in MPEP 2106.05(f), which does not provide integration into a practical application.
Claim 15 recites that the steps are performed using “a processor”, which is a high-level recitation of generic computer components and represents mere instructions to apply on a computer as in MPEP 2106.05(f), which does not provide integration into a practical application.
The claims recite the following additional elements:
Receiving a query from a query-based application that is to perform a query-based task in response to the query;
The limitation recites an insignificant extra solution activity as retrieval of data (ie. Mere data gathering) such as ‘obtaining information’ as identified in MPEP 2106.05(g) and does not provide integration into a practical application.
Creating a semantic cache, via machine learning, based on training data
The limitation of creating a semantic cache via machine learning is understood as a high-level recitation of generic computer components and represents mere instructions to apply on a computer as in MPEP 2106.05(f), which does not provide integration into a practical application.
The training data with a plurality of training samples each of which includes a past query and information representing an intent associated with the past query,
The limitation of the contents of the cache is further understood as reciting an insignificant extra-solution activity as selecting a particular type of data being included in the training data as identified in MPEP 2106.05(g) and does not provide integration into a practical application.
wherein the semantic cache includes a plurality of triplets, each of which includes a past query, an intent related to the past query, and a topic associated with the past query.
The limitation of the contents of the cache is further understood as reciting an insignificant extra-solution activity as selecting a particular type of data being stored in the cache as identified in MPEP 2106.05(g) and does not provide integration into a practical application.
Viewing the additional limitations together and the claim as a whole, nothing provides integration into a practical application.
STEP 2B:
The conclusions for the mere implementation using a computer are carried over and does not provide significantly more.
With respect to “Receiving a query from a query-based application that is to perform a query-based tasks in response to the query;” identified as insignificant extra-solution activity above this is also WURC as court-identified see MPEP 2106.05(d)(II)(i).
Looking at the claim as a whole does not change this conclusion and the claim is ineligible.
As per Dependent Claims 2-7, 9-14, and 16-20,
STEP 1:In accordance with Step 1 of the eligibility inquiry (as explained in MPEP 2106), the claimed method (claims 1-7), machine readable and non-transitory medium (claims 8-14), and system (claims 15-20) are directed to one of the eligible categories of subject matter and therefore satisfies Step 1.
STEP 2A Prong One:The dependent claims 2-7, 9-14, and 16-20 recite the following limitations directed to an abstract idea:
The limitation(s) of Dependent Claims 4, 11, and 17 includes the step(s) of:
Wherein creating the semantic cache comprises:
Generating the plurality of triplets based on the relevant information by, with respect to each of the past queries,
The limitation recites a mental process of observation, evaluation, judgement, and/or opinion capable of being performed by the human mind, or by a human using a pen and paper, by observing and evaluating the relevant information, and based on the observation and evaluation, generating a plurality of triplets using the relevant information.
Estimating intent and topic of the past query by analyzing, with respect to the past query, content of the online document, and
The limitation recites a mental process of observation, evaluation, judgement, and/or opinion capable of being performed by the human mind, or by a human using a pen and paper, by observing and evaluating the past query and content of the online document, and based on the observation and evaluation, making a judgement and/or opinion of an estimate of the intent and topic of the past query.
Generating a triplet including the past query, the intent, and the topic estimated based on the online document with respect to the past query.
The limitation recites a mental process of observation, evaluation, judgement, and/or opinion capable of being performed by the human mind, or by a human using a pen and paper, by observing and evaluating the pat query, the intent, and the topic estimated based on the online document with respect to the past query, and based on the observation and evaluation, making a judgement and/or opinion to organize the past query, the intent, and the topic into a triplet.
The limitation(s) of Dependent Claims 5, 12, and 18 includes the step(s) of:
Wherein the estimating the intent and the topic of the query comprises:
Determining whether at least one of the plurality of triplets exists in the semantic cache, each having its past query similar to the query;
The limitation recites a mental process of observation, evaluation, judgement, and/or opinion capable of being performed by the human mind, or by a human using a pen and paper, by observing and evaluating the plurality of triplets in the semantic cache and the query, and based on the observation and evaluation, making a judgement and/or opinion that at least one of the plurality of triplets, each having its past query similar to the query, exists.
If the at least one triplet exists,
Selecting, from the at least one triplet a matching triplet having its past query best match with the query,
The limitation recites a mental process of observation, evaluation, judgement, and/or opinion capable of being performed by the human mind, or by a human using a pen and paper, by observing and evaluating that at least one triplet exists, observing and evaluating the at least one triplet and the query, and based on the observation and evaluation, making a judgement and/or opinion of a matching triplet having its past query best match with the query.
If the at least one triplet does not exist, predicting, using previously trained intent/topic prediction models, the estimated intent and the estimated topic based on the query.
The limitation recites a mathematical concept of executing a mathematical formula/function (previously trained intent/topic prediction model) in the form of taking as input the query, and outputting the estimated intent and the estimated topic.
The limitation(s) of Dependent Claims 6, 13, and 19 includes the step(s) of:
Wherein the step of determining comprises:
Calculating a word overlap between the query and a past query of each of the plurality of triplets;
The limitation recites a mathematical concept of executing a mathematical formula/function that takes as input the query and past query of each of the plurality of triplets, and performs a word overlap calculation between the query and each past query.
Selecting candidate triplets based on the calculated word overlap, wherein the word overlap between the past query in each of the selected candidate triplets and the query satisfies a first criterion;
The limitation recites a mental process of observation, evaluation, judgement, and/or opinion capable of being performed by the human mind, or by a human using a pen and paper, by observing and evaluating the calculated word for each of the past query, the query, and a first criterion, and based on the observation and evaluation, making a judgement and/or opinion of selected candidate triplets and the query that satisfy the first criterion.
Computing, for each of the candidate triplets, a similarity metric between the past query therein and the query; and
The limitation recites a mathematical concept of executing a mathematical formula/function that, for each candidate triplet, takes as input the past query and the query, and outputs a similarity metric between the past query and the query.
Selecting the at least one triplet, wherein the similarity metric of each of the at least one triplet satisfies a second criterion.
The limitation recites a mental process of observation, evaluation, judgement, and/or opinion capable of being performed by the human mind, or by a human using a pen and paper, by observing and evaluating the calculated similarity metric between each of the at least one triplet and the query and a second criterion, and based on the observation and evaluation, making a judgement and/or opinion of selecting at least one triplet that has a similarity metric that satisfies the second criterion.
The limitation(s) of Dependent Claims 7, 14, and 20 includes the step(s) of:
Generating, via machine learning based on the retrieved plurality of triplets in the semantic cache, the intent/topic prediction models for predicting, based on a given query, an intent, and a topic of the given query.
The limitation recites a mathematical concept of executing a mathematical formula/function that executes a machine learning formula/function using the input of triplets in the semantic cache, and outputs/generates the intent/topic prediction models for predicting, based on a given query, an intent, and a topic of the given query.
STEP 2A Prong Two:The claim(s) recite the following additional elements:
The limitation(s) of Dependent Claims 2, 9, and 16 includes the step(s) of:
Wherein the query is a long-tail query.
The limitation recites an insignificant extra-solution activity as selecting a particular type of query to be used as the query as identified in MPEP 2106.05(g) and does not provide integration into a practical application.
The limitation(s) of Dependent Claims 3 and 10 includes the step(s) of:
Wherein the training data is from one or more data sources.
The limitation recites an insignificant extra-solution activity as selecting the particular type of data source(s) (one or more data sources) the training data comes from as identified in MPEP 2106.05(g) and does not provide integration into a practical application.
The limitation(s) of Dependent Claims 4, 11, and 17 includes the step(s) of:
Wherein each training sample comprises the past query, a title of an online document based on the past query, and a uniform resource locator (URL) pointing to the online document, and
The limitation recites an insignificant extra-solution activity as selecting the particular type of data (past query, title of an online document based on the past query, and a URL pointing to the online document) used to represent each training sample as identified in MPEP 2106.05(g) and does not provide integration into a practical application.
Retrieving the online document based on the URL,
The limitation recites an insignificant extra solution activity as retrieval of data (ie. Mere data gathering) such as ‘obtaining information’ as identified in MPEP 2106.05(g) and does not provide integration into a practical application.
The limitation(s) of Dependent Claims 5, 12, and 18 includes the step(s) of:
Providing the intent and the topic included the matching triplet as the estimated intent and the estimated topic of the query;
The limitation recites an insignificant extra solution activity as sending/receiving data (ie. Mere data gathering) such as ‘obtaining information’ as identified in MPEP 2106.05(g) and does not provide integration into a practical application.
The limitation(s) of Dependent Claims 7, 14, and 20 includes the step(s) of:
Retrieving the plurality of triplets in the semantic cache;
The limitation recites an insignificant extra solution activity as retrieval of data (ie. Mere data gathering) such as ‘obtaining information’ as identified in MPEP 2106.05(g) and does not provide integration into a practical application.
Viewing the additional limitations together and the claim as a whole, nothing provides integration into a practical application.
STEP 2B:
The conclusions for the mere implementation using a computer are carried over and does not provide significantly more.
With respect to Claims 2, 9, and 16 reciting “Wherein the query is along-tail query.” identified as insignificant extra-solution activity above this is also WURC when claimed in a merely generic manner as court-identified see MPEP 2106.05(d)(II)(iv).
With respect to Claim 3 and 10 reciting “Wherein the training data is from one or more data sources.” identified as insignificant extra-solution activity above this is also WURC when claimed in a merely generic manner as court-identified see MPEP 2106.05(d)(II)(iv).
With respect to Claims 4, 11, and 17 reciting “Wherein each training sample comprises the past query, a title of an online document based on the past query, and a uniform resource locator (URL) pointing to the online document,” identified as insignificant extra-solution activity above this is also WURC when claimed in a merely generic manner as court-identified see MPEP 2106.05(d)(II)(iv).
With respect to Claims 4, 11, and 17 reciting “Retrieving the online document based on the URL,” identified as insignificant extra-solution activity above this is also WURC when claimed in a merely generic manner as court-identified see MPEP 2106.05(d)(II)(i).
With respect to Claims 5, 12, and 18 reciting “Providing the intent and the topic included the matching triplet as the estimated intent and the estimated topic of the query;” identified as insignificant extra-solution activity above this is also WURC when claimed in a merely generic manner as court-identified see MPEP 2106.05(d)(II)(i).
With respect to Claims 7, 14, and 20 reciting “Retrieving the plurality of triplets in the semantic cache;” identified as insignificant extra-solution activity above this is also WURC when claimed in a merely generic manner as court-identified see MPEP 2106.05(d)(II)(i).
Looking at the claim as a whole does not change this conclusion and the claim is ineligible.
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.
Claim(s) 1-4, 8-11, and 15-17 are rejected under 35 U.S.C. 103 as being unpatentable over Bakir et al. (U.S. Patent No. 9,830,391, hereinafter referred to as Bakir) and further in view of Gruber et al. (U.S. Pre-Grant Publication No. 2025/0173519, hereinafter referred to as Gruber).
Regarding Claim 1:
Bakir teaches a method, comprising:
a storage including a plurality of triplets, each of which includes a past query, an intent related to the past query, and a topic associated with the past query;
Bakir teaches “if the query evaluator 202 does determine to contextually modify the received query, then the textural resource processor 204 determines modification data for the query (310). Appropriate modification data are data that can be used to generate modified queries from the received query. For example, the non-textual data processor 204 may determine, from the non-textual data, entity text that describes entities. As used herein, entities are topics of discourse, concepts or things that can be referred to by a text fragment, e.g., a term or phrase, and are distinguishable from one another, e.g., based on context” (Col. 6 Lines 19-39) and “Topics can also be used as modification data. For example, an image or video may have been categorized as belonging to one or more topics, and each topic name can be used as modification data.” (Col. 6 Line 64 – Col. 7 Line 8)
Bakir further teaches “When performing a context modification, the system determines, based on the non-textual resource, modification data for the query, and generates a set of modified queries based on the query and the modification data. A variety of modification data types and query modification algorithms can be used. The system then scores the modified queries according to one or more scoring criteria. The score for each query may be based on the results of a search operation performed for each query, the similarity of the query to other, previously received queries, and other criteria.” (Col. 2 Lines 39-52) and “the queries 109 submitted from user devices 106 are stored in query logs 114….The query logs 114 and the click data 116 define search history data 117 that include data from and related to previous search requests associated with unique identifiers. The click data define actions taken responsive to search results provided by the search system 110. The query logs 114 and click logs 116 can be used to map queries submitted by the user devices to web pages that were identified in search results and the actions taken by users (i.e., that data are associated with the identifiers from the search requests so that a search history for each identifier can be accessed). The click logs 116 and query logs 114 can thus be used by the search system to the sequence of queries submitted by the user devices, the actions taken in response to the queries, and how often the queries are submitted.” (Col. 4 Lines 5-35).
Bakir further teaches representing queries and other resources with “feature vectors” (Col. 3 Lines 49 - 63) where the triplet herein is merely understood in the art as a vector of size 3.
Receiving a query from a query-based application that is to perform a query-based task in response to the query;
Bakir teaches “the query evaluator 202 receives a query (302). For example, the search system 110 receives the query Q from the user device 106.” (Col. 5 Lines 10-14) and “The candidate modified query with the highest score relative to other candidate modification queries is selected by the candidate scorer 208, and is provided to the search system front end 210. The search system front end 210 then provides search results responsive to the modified query (316). Alternatively, if the modified query triggers an action, then data is sent to the user device that causes the user device to perform the action.” (Col. 8 Lines 60-67).
Estimating an intent and a topic associated with the query based on the stored plurality of triplets;
Bakir teaches “if the query evaluator 202 does determine to contextually modify the received query, then the textural resource processor 204 determines modification data for the query (310). Appropriate modification data are data that can be used to generate modified queries from the received query. For example, the non-textual data processor 204 may determine, from the non-textual data, entity text that describes entities. As used herein, entities are topics of discourse, concepts or things that can be referred to by a text fragment, e.g., a term or phrase, and are distinguishable from one another, e.g., based on context” (Col. 6 Lines 19-39) and “Topics can also be used as modification data. For example, an image or video may have been categorized as belonging to one or more topics, and each topic name can be used as modification data.” (Col. 6 Line 64 – Col. 7 Line 8)
Bakir further teaches “When performing a context modification, the system determines, based on the non-textual resource, modification data for the query, and generates a set of modified queries based on the query and the modification data. A variety of modification data types and query modification algorithms can be used. The system then scores the modified queries according to one or more scoring criteria. The score for each query may be based on the results of a search operation performed for each query, the similarity of the query to other, previously received queries, and other criteria.” (Col. 2 Lines 39-52)
Bakir further teaches representing queries and other resources with “feature vectors” (Col. 3 Lines 49 - 63) where the triplet herein is merely understood in the art as a vector of size 3.
Providing the estimated intent and the estimated topic to the query-based application to facilitate execution of the query-based task in accordance with both the query and the estimated intent and the topic of the query associated therewith.
Bakir teaches “The candidate modified query with the highest score relative to other candidate modification queries is selected by the candidate scorer 208, and is provided to the search system front end 210. The search system front end 210 then provides search results responsive to the modified query (316). Alternatively, if the modified query triggers an action, then data is sent to the user device that causes the user device to perform the action.” (Col. 8 Lines 60-67).
Bakir explicitly teaches all of the elements of the claimed invention as recited above except:
Creating a semantic cache, via machine learning, based on training data with a plurality of training samples each of which includes a past query and information representing an intent associated with the past query,
Estimating an intent and a topic associated with the query based on the semantic cache;
However, in the related field of endeavor of task flow identification based on user intent, Gruber in combination with Bakir teaches:
Creating a semantic cache, via machine learning, based on training data with a plurality of training samples each of which includes a past query and information representing an intent associated with the past query,
Gruber teaches creating a semantic cache by teaching maintaining “subsets and/or portions of these components locally, to improve responsiveness and reduce dependence on network communications. Such subsets and/or portions can be maintained and updated according to well known cache management techniques. Such subsets and/or portions include, for example: subset of vocabulary 1058a; subset of library of language pattern recognizers 1060a; cache of short term personal memory 1052a; cache of long term personal memory 1054a.” (Para. [0105] – [0109]).Gruber further teaches using machine learning by teaching “unlike assistant technology that attempts to implement a general-purpose artificial intelligence system, the embodiments described herein may apply the multiple sources of constraints to reduce the number of solutions to a more tractable size.” (Para. [0127]). It is noted that the claims do not recite how machine learning is used to create a cache or its contents based on the training data.Bakir teaches past data with a plurality of samples including a past query and information representing an intent associated with the past query by teaching maintaining relevant information including past queries, links to online documents selected from search results obtained from past queries, and titles of online documents by teaching “The query logs 114 and the click data 116 define search history data 117 that include data from and related to previous search requests associated with unique identifiers.” (Col. 4 Lines ) where “the score for each query may be based on the results of a search operation performed for each query, the similarity of the query to other, previously received queries, and other criteria” (Col. 2 Lines 39-52) and “an example search result can include a web page title, a snippet of text extracted from the web page, and the URL of the web page” (Col. 3 Lines 33-48).
Estimating an intent and a topic associated with the query based on the semantic cache;
Gruber teaches maintaining “subsets and/or portions of these components locally, to improve responsiveness and reduce dependence on network communications. Such subsets and/or portions can be maintained and updated according to well known cache management techniques. Such subsets and/or portions include, for example: subset of vocabulary 1058a; subset of library of language pattern recognizers 1060a; cache of short term personal memory 1052a; cache of long term personal memory 1054a.” (Para. [0105] – [0109]) thereby teaching a semantic cache for storing a ” subset of vocabulary 1058a; subset of library of language pattern recognizers 1060a; cache of short term personal memory 1052a; cache of long term personal memory 1054a.” with the advantage of “to improve responsiveness and reduce dependence on network communications”.
Thus, it would have been obvious to one of ordinary skill in the art, having the teachings of Gruber and Bakir at the time that the claimed invention was effectively filed, to have modified the systems and methods for modifying queries, as taught by Bakir, with the caching of a subset of data, as taught by Gruber.
One would have been motivated to make such combination because Bakir teaches “search history data 117” comprising query logs 114 and selection logs 116 (Fig. 1), where the search history data 117 is separate from the contextual query modifier and Gruber teaches maintaining “subsets and/or portions of these components locally, to improve responsiveness and reduce dependence on network communications” (Para. [0105] – [0109]) which would have been obvious to a person having ordinary skill in the art to incorporate to the contextual query modifier 120 taught by Bakir to improve responsiveness and reduce dependence on network communications.
Regarding Claim 2:
Gruber and Bakir further teach:
Wherein the query is a long-tail query.
Bakir teaches the query as being able to be a long-tail query by teaching “query logs can be accessed to determine a similarity of a modified query (e.g., the text of the query and the entity text) to other prior queries stored in a query log. Prior queries for which the similarity measure meets a threshold similarity value may be selected as a candidate query. This type of candidate generation can optionally filter out “long tail” queries if the candidate query is a query that is rarely received, e.g., if the candidate query does not meet a minimum frequency threshold or instance count.” (Col. 8 Lines 1-10).
Regarding Claim 3:
Gruber and Bakir further teach:
The training data is from one or more data sources.
Gruber teaches maintaining “subsets and/or portions of these components locally, to improve responsiveness and reduce dependence on network communications. Such subsets and/or portions can be maintained and updated according to well known cache management techniques. Such subsets and/or portions include, for example: subset of vocabulary 1058a; subset of library of language pattern recognizers 1060a; cache of short term personal memory 1052a; cache of long term personal memory 1054a.” (Para. [0105] – [0109]).
Regarding Claim 4:
Gruber and Bakir further teach:
Wherein each training sample comprise the past query, a title of an online document based on the past query, and a uniform resource locator (URL) pointing to the online document, and
Bakir teaches maintaining relevant information including past queries, links to online documents selected from search results obtained from past queries, and titles of online documents by teaching “The query logs 114 and the click data 116 define search history data 117 that include data from and related to previous search requests associated with unique identifiers.” (Col. 4 Lines ) where “the score for each query may be based on the results of a search operation performed for each query, the similarity of the query to other, previously received queries, and other criteria” (Col. 2 Lines 39-52) and “an example search result can include a web page title, a snippet of text extracted from the web page, and the URL of the web page” (Col. 3 Lines 33-48)
Wherein creating the semantic cache comprises:
Generating the plurality of triplets based on the relevant information by, with respect to each of the past queries,
Bakir teaches “the queries 109 submitted from user devices 106 are stored in query logs 114. Click data for the queries and the web pages referenced by the search results are stored in click logs 116” (Col. 4 Lines 5-21) and representing queries and other resources with “feature vectors” (Col. 3 Lines 49 - 63) where the triplet herein is merely understood in the art as a vector of size 3.
Retrieving the online document based on the URL,
Bakir teaches, “the search system 110 will conduct a search based on the query, and the query evaluator 202 will evaluate the results” and “a quality of the top-scoring resources identified as being responsive to the query; a topicality threshold that indicates that the top-scoring resources have a dominant intent of at least one topic, etc.” (Col. 6 Lines 9-18). Bakir further teaches “an example search result can include a web page title, a snippet of text extracted from the web page, and the URL of the web page” (Col. 3 Lines 33-48) where URLs are understood as resource locators for locating and retrieving information.
Estimating intent and topic of the past query by analyzing, with respect to the past query, content of the online document, and
Bakir teaches, “the search system 110 will conduct a search based on the query, and the query evaluator 202 will evaluate the results” and “a quality of the top-scoring resources identified as being responsive to the query; a topicality threshold that indicates that the top-scoring resources have a dominant intent of at least one topic, etc.” (Col. 6 Lines 9-18).
Generating a triplet including the past query, the intent, and the topic estimated based on the online document with respect to the past query.
Bakir teaches “the queries 109 submitted from user devices 106 are stored in query logs 114. Click data for the queries and the web pages referenced by the search results are stored in click logs 116” (Col. 4 Lines 5-21) and representing queries and other resources with “feature vectors” (Col. 3 Lines 49 - 63) where the triplet herein is merely understood in the art as a vector of size 3.
Regarding Claim 8:
Some of the limitations herein are similar to some or all of the limitations as recited in Claim 1.
Gruber and Bakir further teach:
A machine-readable and non-transitory medium having information recorded thereon, wherein the information, when read by the machine, causes the machine to perform steps (Bakir – Col. 11 Lines 9-31 & Claim 19).
Regarding Claim 9:
All of the limitations herein are similar to some or all of the limitations as recited in Claim 2.
Regarding Claim 10:
All of the limitations herein are similar to some or all of the limitations as recited in Claim 3.
Regarding Claim 11:
All of the limitations herein are similar to some or all of the limitations as recited in Claim 4.
Regarding Claim 15:
Some of the limitations herein are similar to some or all of the limitations as recited in Claim 1.
Gruber and Bakir further teach a system comprising:
A processor (Bakir – Col. 11 Lines 36-52);
Regarding Claim 16:
All of the limitations herein are similar to some or all of the limitations as recited in Claim 2.
Regarding Claim 17:
All of the limitations herein are similar to some or all of the limitations as recited in Claim 4.
Claim(s) 5-7, 12-14, and 18-20 are rejected under 35 U.S.C. 103 as being unpatentable over Zhou, Gruber, and Bakir, and further in view of Daianu et al. (U.S. Patent No. 11,650,996, hereinafter referred to as Daianu) and Zhou et al. (U.S. Pre-Grant Publication No. 2025/0342218, hereinafter referred to as Zhou).
Regarding Claim 5:
Gruber and Bakir further teach:
Wherein the estimating the intent and the topic of the query comprises:
Determining whether at least one of the plurality of triplets exists in the semantic cache, each having its past query similar to the query;
Bakir teaches “The score for each query may be based on the results of a search operation performed for each query, the similarity of the query to other, previously received queries, and other criteria. A modified query with a highest score relative to the other modified queries is selected” (Col. 2 Lines 39-52).
If the at least one triplet exists,
Selecting, from the at least one triplet a matching triplet having its past query best match with the query,
Bakir teaches “all candidate queries are provided to the candidate scorer 208. In other implementations, the candidate generator 206 (or the candidate scorer 208) can implement a filtering process to filter out queries that include latent or patent type mismatches of terms. The mismatches can be detected using language models, attribute and entity maps, and the like. For example, the query [What is the telephone number June 1] has a patent mismatch between the attribute “telephone number” and the entity “June 1,” and thus is removed from the scoring process.” (Col. 7 Lines 48-58)
Providing the intent and the topic included the matching triplet as the estimated intent and the estimated topic of the query; and
Bakir teaches “The candidate modified query with the highest score relative to other candidate modification queries is selected by the candidate scorer 208, and is provided to the search system front end 210.” (Col. 8 Lines 60-67) thereby teaching providing the intent and topic features related to the previous query, used in the candidate modified query, as the modified query for generating search results.
Gruber and Bakir explicitly teach all of the elements of the claimed invention as recited above except:
If the at least one triplet does not exist, predicting, using previously trained intent/topic prediction models, the estimated intent and the estimated topic based on the query.
However, in the related field of endeavor of task flow identification based on user intent, Gruber teaches:
predicting, using previously trained intent/topic prediction models, the estimated intent and the estimated topic based on the query.
Daianu teaches “Intent system 130 may train intent model 132 using a variety of methods. In general, training a machine learning model can be classified into one of three methods: supervised learning, semi-supervised learning and unsupervised learning.” (Col. 10 Lines 23-27).
Daianu further teaches “intent model 132 may be trained by intent system 130 but execute on message routing server 120. Intent model 132 may be a variety of different machine learning models, including logistic regression models, decision trees, random forests, and deep neural networks. Intent model 132 accepts a vector representing text as input (in this case, query vector 124) as input and outputs a determined intent of the text represented by the vector (shown as query intent 134).” (Col. 4 Lines 5-21).
Thus, it would have been obvious to one of ordinary skill in the art, having the teachings of Daianu, Gruber, and Bakir at the time that the claimed invention was effectively filed, to have modified the systems and methods for modifying queries, as taught by Bakir, and the caching of a subset of data, as taught by Gruber, with the stop word module for removing stop words form the text query, as taught by Daianu.
One would have been motivated to make such combination because Daianu teaches “Stop words may be removed before natural language processing in order to improve processing time and efficiency, because they are common words that convey little meaning, but may complicate searching for phrases” (Col. 7 Lines 19-35) and it would have been obvious to a person having ordinary skill in the art that improving processing time and efficiency would reduce the cost and energy required for performing the processing.
Daianu, Gruber, and Bakir explicitly teach all of the elements of the claimed invention as recited above except:
If the at least one triplet does not exist, predicting, using previously trained intent/topic prediction models, the estimated intent and the estimated topic based on the query.
However, in the related field of endeavor of comparing query information to cached data and generating query information, Zhou in combination with Gruber and Bakir teaches:
If the at least one triplet does not exist, predicting, using previously trained intent/topic prediction models, the estimated intent and the estimated topic based on the query.
Zhou teaches “determine whether the input search query accessed by search query input component 230 is semantically similar to previous search query that were used to generate summaries and cached by summary caching component 280…determine whether an input search query is above a threshold similarity to a previous input search query with a corresponding cached generated summary….[and] initiate generation of a summary via search result summary engine 210 when there are no semantically similar previous search query that were used to generate summaries and cached by summary caching component 280.” (Para. [0046]).
Thus, it would have been obvious to one of ordinary skill in the art, having the teachings of Zhou, Daianu, Gruber, and Bakir at the time that the claimed invention was effectively filed, to have modified the systems and methods for modifying queries, as taught by Bakir, the caching of a subset of data, as taught by Gruber, and the stop word module for removing stop words form the text query, as taught by Daianu, with the checking for already generated query-related content and new generation of query-related content if it does not already exist, as taught by Zhou.
One would have been motivated to make such combination because Zhou teaches checking a cache of previously generated data for similarity to another input search query “facilitates improved search engine processing and response times…in response to input search queries” and “computer resources including computer processing, memory, and network communication bandwidth are conserved because the cached generated summaries are accessed and presented instead of calling a language model to generate a new generated summary” (Para. [0027]) but also dynamically allows for the system to “initiate generation of [new information for the search query] when there are no semantically similar previous search query” (Para. [0046]).
Regarding Claim 6:
Zhou, Daianu, Gruber, and Bakir further teach:
Wherein the step of determining comprises:
Calculating a word overlap between the query and a past query of each of the plurality of triplets;
Daianu teaches “intent model 132 analyzes vectors to identify keywords in the vectors. For example, to identify an intent such as “printing” intent model 132 may determine that a vector is closely associated with other vectors known to intent model 132 to correspond to the intent “printing,” using vector comparison techniques.” (Col. 9 Lines 46-62)
Bakir teaches “The score for each query may be based on the similarity of the query to other, previously received queries, and other criteria” (Col. 2 Lines 39-52).Therefore, Daianu in combination with Bakir teaches calculating the word overlap using vectors to compare similarity of a query to other previously received queries.
Selecting candidate triplets based on the calculated word overlap, wherein the word overlap between the past query in each of the selected candidate triplets and the query satisfies a first criterion;
Daianu teaches “intent model 132 analyzes vectors to identify keywords in the vectors. For example, to identify an intent such as “printing” intent model 132 may determine that a vector is closely associated with other vectors known to intent model 132 to correspond to the intent “printing,” using vector comparison techniques.” (Col. 9 Lines 46-62) where “a query vector representing the text “can I file 1099MISC” may be assigned a relatively high score for the intent class “1099MISC.”” (Col. 9 Line 63 – Col. 10 Line 8).
Computing, for each of the candidate triplets, a similarity metric between the past query therein and the query; and
Bakir teaches “The score for each query may be based on the similarity of the query to other, previously received queries, and other criteria” (Col. 2 Lines 39-52).
Selecting the at least one triplet, wherein the similarity metric of each of the at least one triplet satisfies a second criterion.
Bakir teaches “A modified query with a highest score relative to the other modified queries is selected, and search results responsive to the selected modified query are provided to the user device from which the original query was received.” (Col. 2 Lines 39-52).
Regarding Claim 7:
Zhou, Daianu, Gruber, and Bakir further teach:
Retrieving the plurality of triplets in the semantic cache;
Daianu teaches “query vector 124 has been prepared for use in training a machine learning model, and as a result, query vector 124 includes intent class 232 and complexity class 234. To be used in training, a query vector is labeled with an intent class (for use in training an intent model) and a complexity class (for use in training a complexity model).” (Col. 9 Lines 8-20).
Generating, via machine learning based on the retrieved plurality of triples in the semantic cache, the intent/topic prediction models for predicting, based on a given query, an intent, and a topic of the given query.
Daianu teaches “Intent system 130 may train intent model 132 using a variety of methods. In general, training a machine learning model can be classified into one of three methods: supervised learning, semi-supervised learning and unsupervised learning.” (Col. 10 Lines 23-27).
Daianu further teaches “intent model 132 may be trained by intent system 130 but execute on message routing server 120. Intent model 132 may be a variety of different machine learning models, including logistic regression models, decision trees, random forests, and deep neural networks. Intent model 132 accepts a vector representing text as input (in this case, query vector 124) as input and outputs a determined intent of the text represented by the vector (shown as query intent 134).” (Col. 4 Lines 5-21).
Regarding Claim 12:
All of the limitations herein are similar to some or all of the limitations as recited in Claim 5.
Regarding Claim 13:
All of the limitations herein are similar to some or all of the limitations as recited in Claim 6.
Regarding Claim 14:
All of the limitations herein are similar to some or all of the limitations as recited in Claim 7.
Regarding Claim 18:
All of the limitations herein are similar to some or all of the limitations as recited in Claim 5.
Regarding Claim 19:
All of the limitations herein are similar to some or all of the limitations as recited in Claim 6.
Regarding Claim 20:
All of the limitations herein are similar to some or all of the limitations as recited in Claim 7.
Response to Amendment
Applicant’s Amendments, filed on 4/16/2026, are acknowledged and accepted.
In light of the Amendments and Remarks filed on 4/16/2026, the objections to claims 1, 8, 13, and 15 have been withdrawn.
In light of the Amendments and Remarks filed on 4/16/2026, the 112(b) rejection of claims 5-7, 12-14, and 18-20 has been withdrawn.
Response to Arguments
On pages 13-14 of the Remarks filed on 4/16/2026, Applicant states that “Amended claim 1 recites ‘creating a semantic cache, via machine learning, based on training data with a plurality of training samples each of which includes a past query and information representing an intent associated with the past query, wherein the semantic cache includes a plurality of triplets, each of which includes a past query, an intent related to the past query, and a topic associated with the past query.’ The above-quoted claim features cannot be performed by a human mind or a human with pen/paper at least because a human mind cannot create a cache via machine learning. Thus, claim 1 does not fall within the "mental processes" grouping of abstract ideas. Also, the above-quoted claim features are not directed to mathematical concepts. Accordingly, claim 1 does not fall into any of the abstract ideas exceptions provided by the MPEP, and thus claim 1 is patent eligible under Prong One of the Step 2A Analysis.”
Applicant’s argument is moot because creating a cache was never stated as being a mental process or directed to a mathematical concept. However, the amended limitation being argued has been addressed in the rejection above.
On pages 14-16 of the Remarks filed on 4/16/2026, Applicant argues that “[amended] claim 1 is patent eligible because the claimed concepts are integrated into a practical application” because “[amended] Claim 1 provides an improvement in the technical field of information processing. Para. [0001] as filed. In this technical field, ‘[t]here is a need for an approach capable of better understanding a query but also more accurately estimating the intent associated with the query in order to uncover content relevant to the query according to the intent so as to enable enhanced application performance based on the searched content.’ (Para. [0005]).” and “As [amended] claim 1 requires creating a machine-learned semantic cache that enables efficient on- the-fly estimation of an intent/topic associated with a long-tail query to be utilized in a query-based application, it integrates any alleged abstract idea into a practical application.”Applicant’s argument is not convincing because it appears to be tying an argued practical application (efficient on-the-fly estimation of an intent/topic associated with a long-tail query to be utilized in a query-based application) to a limitation that does not actually perform this step (creating a machine-learned semantic cache) and therefore cannot integrate the claims into the argued practical application.
On pages 16-17 of the Remarks filed on 4/16/2026, Applicant argues that “[amended] claim 1 amounts to significantly more than the judicial exception” because “In addition to failing to consider Applicant's claims as an ordered combination and as a whole, the Office Action has improperly analyzed the claims without considering the "additional element(s)" in combination with the non-additional elements. As a result, the Office Action has also incorrectly and improperly identified that the alleged "additional elements" do not amount to significantly more than the alleged judicial exception.”Applicant’s argument is not convincing because the claims were both considered as an ordered combination and as a whole, including the additional and non-additional elements, in the office action dated 1/16/2026.
On pages 19-20 of the Remarks filed on 4/16/2026, Applicant argues that “none of the cited references teaches or suggests the [amended] claim features” related to “creating a semantic cache, via machine learning, based on training data” because
“Bakir appears to teach modifying queries based on topics and providing search results based on modified queries, wherein the search results may include a web page title, a snippet of text extracted from the web page, and the URL of the webpage. However, Bakir does not teach or suggest machine learning a cache "includ[ing] a plurality of triplets, each of which includes a past query, an intent related to the past query, and a topic associated with the past query" based on "training data with a plurality of training samples each of which includes a past query and information representing an intent associated with the past query," as required by [amended] claim 1.”
“Gruber does not teach or suggest machine learning a cache "includ[ing] a plurality of triplets, each of which includes a past query, an intent related to the past query, and a topic associated with the past query" based on "training data with a plurality of training samples each of which includes a past query and information representing an intent associated with the past query," as required by [amended] claim 1.” and
“Daianu still does not teach or suggest machine learning a cache "includ[ing] a plurality of triplets, each of which includes a past query, an intent related to the past query, and a topic associated with the past query" based on "training data with a plurality of training samples each of which includes a past query and information representing an intent associated with the past query as required by [amended] claim 1.”
Applicant’s argument is not convincing because upon further review, it is the combination of Bakir and Gruber that teaches the argued amended limitation, which is addressed in full in the rejection above.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Chiang et al. (U.S. Pre-Grant Publication No. 2025/0173519) teaches “the system 200 receives an input query 202 and multi-turn query data 204 as input. In some instances, the input query 202 may be indicative of a user's search query” (Para. [0042]) and “the input query 202 and multi-turn query data 204 may be processed with a generative language model 206 (e.g., large language model) to determine an intent of the input query 202 based on the multi-turn query data 204.” where (Para. [0042]) where “The intent graph 308 can include a plurality of nodes and edges and may include a graphical representation that identifies similar queries with similar intents.” and “The edges between the embedding cluster(s) 309 may link clusters 309 with closely related intents and/or other attributes (e.g., topics, subjects, etc.)” (Para. [0046]).
Tremblay et al. (U.S. Pre-Grant Publication No. 2018/0349500) teaches generating search results responsive to receipt of a query. More specifically, the query is mapped to a topic in response to receipt of a query, and social media accounts that have been labeled as being knowledgeable on the topic are identified. Messages in a message feed of the social media account that are germane to the topic are retrieved, and documents referenced (linked) in the retrieved messages are identified. These documents are positioned in a ranked list based upon the documents being referenced in the messages.
Aghajanyan et al. (U.S. Pre-Grant Publication No. 2021/0117624) teaches receiving a user input comprising a natural-language utterance by an assistant xbot from a client system associated with a user, determining a semantic representation of the user input based on a structural ontology defining a labeling syntax for parsing the natural-language utterance to semantic units comprising actions, objects, and attributes, wherein the semantic representation embeds at least one object within at least one action and declares at least one attribute of the embedded object to be acted upon, sending a request based on the semantic representation to an agent for executing a task corresponding to the user input, receiving results of the executed task mapped to a structure determined by the structural ontology from the agent, and sending from the assistant xbot to the client system instructions for presenting a response based on the results of the executed task.
Shukla et al. (U.S. Patent No. 9,128,945) teaches generating or using augmentation queries. In one aspect, a first query stored in a query log is identified and a quality signal related to the performance of the first query is compared to a performance threshold. The first query is stored in an augmentation query data store if the quality signal indicates that the first query exceeds a performance threshold.
Liu et al. (U.S. Pre-Grant Publication No. 2023/0130232) teaches predicting responses to DNS queries. The method includes receiving a DNS query comprising a subdomain portion and a root domain portion from a client device, determining whether to obtain target address information corresponding to the DNS from a predictive cache, in response to determining to obtain the target address information from the predictive cache, obtaining the target address information from the predictive cache, and providing the target address information to the client device.The reference further teaches “the predictive cache is generated using a machine learning process” (Claim 9).
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ROBERT F MAY whose telephone number is (571)272-3195. The examiner can normally be reached Monday-Friday 9:30am to 6pm.
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, Boris Gorney can be reached on 571-270-5626. 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.
/ROBERT F MAY/Examiner, Art Unit 2154 6/10/2026
/BORIS GORNEY/Supervisory Patent Examiner, Art Unit 2154
Prosecution Insights