DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This action is responsive to communications: Amendment filed on 2/23/2026.
Claims 1-20 are pending. Claims 1, 11, and 16 are independent.
The rejection of claims 1-15 under 35 USC § 101 has been withdrawn in view of the amendment.
The previous rejection of claims 1-20 under 35 USC § 102 and 35 USC § 103 have been withdrawn in view of the amendment.
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, 6-8, 10-12, and 16-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Amoateng et al. (US2024/0320257) in view of Cui et al. (US2025/0077940) and Baeuml et al. (US2024/0311402).
In regards to claim 1, Amoateng et al. discloses a computing system comprising:
a processor (Amoateng et al. fig. 6 610 para[0083]); and
computer storage memory having computer-executable instructions stored thereon which (Amoateng et al. fig. 6 650 para[0083]), when executed by the processor, configure the computing system to perform operations comprising:
obtain text associated with a source content (Amoateng et al. para[0074], the ranked content is received);
obtain, as output from the large language model, the alternative example for the user segment (Amoateng et al. para[0058], When the user selects this global pill prompt, the display of the personalized content feed is changed so that only those feed items that correspond to important things that the user has missed this week are displayed in the feed); and
provide, for display via a user interface, the alternative example for the user segment in association with the source content (Amoateng et al. para[0059], When a subset of the feed is displayed in this manner, pills for each of the feed items in the displayed subset of the content feed may be selected as normal, with the display of the response provided in response to the selection of one of the pills in the same manner as is done when the entire feed is provided).
Amoateng et al. does not explicitly disclose identify a source example comprising an entity-context pair from the text of the source content by:
generating a source example prompt to be input into a large language model, the source example prompt including the text associated with the source content and an instruction to generate the source example from the text associated with the source content for use in generating one or more alternative examples in association with one or more user segments, and
obtaining, as output from the large language model, the source example that represents an entity and corresponding context from the text of the source content;
generate an alternative example prompt to be input into the large language model, the alternative example prompt including the source example comprising the entity-context pair generated via the large language model, an indication of a user segment, and an instruction to generate an alternative example for the user segment in accordance with the source example.
However Cui et al. discloses identify a source example comprising an entity-context pair from the text of the source content by:
generating a source example prompt to be input into a large language model, the source example prompt including the text associated with the source content and an instruction to generate the source example from the text associated with the source content for use in generating one or more alternative examples in association with one or more user segments (CUI et a. fig. 4 para[0052], generates initial prompt from user query, instructions and content), and
obtaining, as output from the large language model, the source example that represents an entity and corresponding context from the text of the source content (CUI et al. para[0032], generates answer from initial prompt provided to ML model 118, para[0001], ML models such as large language models);
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the verification method of Cui et al. in order to identify and prevent hallucinations from large language models (Cui et al. para[0002]). Source example: NL based input, additional LLM input
Amoateng et al. does not explicitly disclose generate an alternative example prompt to be input into the large language model, the alternative example prompt including the source example comprising the entity-context pair generated via the large language model, an indication of a user segment, and an instruction to generate an alternative example for the user segment in accordance with the source example.
However Baeuml et al. discloses generate an alternative example prompt to be input into the large language model, the alternative example prompt including the source example comprising the entity-context pair generated via the large language model, an indication of a user segment, and an instruction to generate an alternative example for the user segment in accordance with the source example (Baeuml et al. para[0060], system process NL based input and based on first segment a generates additional LLM output.
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the rendering method of Baeuml et al. in order to reduce latency in generating natural language based output (Baeuml et al. para[0004]).
In regards to claim 6, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the computing system of claim 1, wherein the alternative example is provided for display via the user interface by supplementing a presentation of the source content (Amoateng et al. fig. 4 para[0076], a content feed is displayed to the first user, including causing a plurality of selectable pills to be displayed to the first user).
In regards to claim 7, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the computing system of claim 6, wherein the alternative example is provided for display in response to a user selection of a user segment option representing the user segment (Amoateng et al. fig. 4 para[0076], responsive to the selection of one of the selectable pills, causing the response to the pill prompt that corresponds to the selection to be displayed to the first user.).
In regards to claim 8, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the computing system of claim 1, wherein the alternative example is provided for display via the user interface by modifying the source content to include the alternative example (Amoateng et al. fig. 4 para[0076], a content feed is displayed to the first user, including causing a plurality of selectable pills to be displayed to the first user).
In regards to claim 10, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the computing system of claim 1, wherein the user segment comprises at least one of a topic and a demographic (Amoateng et al. para[0027], The user profile may include a variety of information about a user, including basic information such as name, job title, date of birth, employment information, residence information, or other suitable information. The user profile may also include information such as topics of interest to the user and other suitable information about the user. The user profile consists of key/value pairs).
In regards to claim 11, Amoateng et al. discloses a computer-implemented method comprising:
based on a selection of a particular user segment, providing, for display via a user interface, an alternative example, of the set of alternative examples, that corresponds to the particular user segment (Amoateng et al. para[0059], When a subset of the feed is displayed in this manner, pills for each of the feed items in the displayed subset of the content feed may be selected as normal, with the display of the response provided in response to the selection of one of the pills in the same manner as is done when the entire feed is provided).
Amoateng et al does not explicitly disclose Identifying a source example comprising an entity-context pair from the text of the source content by:
generating a source example prompt to be input into a large language model, the source example prompt including text associated with a source content and an instruction to generate a source example from the text associated with the source content (Amoateng et al. para[0075], a first prompt is provided such that the first prompt includes natural language text instructions) for use in generation one or more alternative examples in association with a set of user segments, and
obtaining, as output from the large language model, the source example that represents an entity and corresponding context from the text;
providing, as input into the large language model, the source example comprising the entity-context pair from the text and the set of user segments for use in generating alternative examples associated with the source content;
obtaining, as output from the large language model, a set of alternative examples generated based on the source example, each alternative example corresponding to a user segment of the set of user segments;
verifying that each alternative example of the set of alternative examples is factually correct using search results associated with the alternative example.
However Cui et al. discloses Identifying a source example comprising an entity-context pair from the text of the source content by:
generating a source example prompt to be input into a large language model, the source example prompt including text associated with a source content and an instruction to generate a source example from the text associated with the source content (Amoateng et al. para[0075], a first prompt is provided such that the first prompt includes natural language text instructions) for use in generation one or more alternative examples in association with a set of user segments (CUI et a. fig. 4 para[0052], generates initial prompt from user query, instructions and content), and
obtaining, as output from the large language model, the source example that represents an entity and corresponding context from the text (CUI et al. para[0032], generates answer from initial prompt provided to ML model 118, para[0001], ML models such as large language models);
obtaining, as output from the large language model, a set of alternative examples generated based on the source example, each alternative example corresponding to a user segment of the set of user segments (Cui para[0034], generates alternative verification prompt from initial answer);
verifying that each alternative example of the set of alternative examples is factually correct using search results associated with the alternative example (Cui et al. para[0053], a “NO” response to the verification prompt is indicative of the potential answer being a hallucination).
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the verification method of Laban et al. in order to identify correct factual errors in personalized summaries (Laban et al. pg163 section1 para2).
Amoateng et al. does not explicitly disclose providing, as input into the large language model, the source example comprising the entity-context pair from the text and the set of user segments for use in generating alternative examples associated with the source content.
However Baeuml et al. discloses providing, as input into the large language model, the source example comprising the entity-context pair from the text and the set of user segments for use in generating alternative examples associated with the source content. (Baeuml et al. para[0060], system process NL based input and based on first segment a generates additional LLM output).
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the rendering method of Baeuml et al. in order to reduce latency in generating natural language based output (Baeuml et al. para[0004]).
In regards to claim 12, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the method of claim 11 further comprising providing, for display via the user interface, a set of user segment options in association with the source content, the set of user segment options including an option of the particular user segment (Amoateng et al. para[0070], When a subset of the feed items is displayed in this manner, the feed items are displayed in the same manner as when the entire feed is displayed, including the personalization for each feed item such as personalized pill prompts for the feed item, the personalized title, the personalized summary, and the personalized picture, but a personalized subset of the entire feed is displayed in response to the selection of the global pill prompt.).
In regards to claim 16, Amoateng et al. discloses one or more computer storage media having computer-executable instructions embodied thereon that, when executed by one or more processors, cause the one or more processors to perform a method, the method comprising:
obtaining, at a large language model, a source example prompt including text associated with a source content and an instruction to generate a source example from the text associated with the source content (Amoateng et al. para[0075], a first prompt is provided such that the first prompt includes natural language text instructions);
based on a particular user segment associated with a user interested in the source content, causing presentation of an alternative example corresponding to the particular user segment (Amoateng et al. para[0059], When a subset of the feed is displayed in this manner, pills for each of the feed items in the displayed subset of the content feed may be selected as normal, with the display of the response provided in response to the selection of one of the pills in the same manner as is done when the entire feed is provided).
Amoateng et al. does not explicitly disclose generating, using the large language model, the source example that includes an entity-context pair from the text.
providing, as input into the large language model, the source example and a set of user segments to generate alternative examples associated with the source content, each alternative example corresponding to a user segment of the set of user segments; and
Cui et al. discloses generating, using the large language model, the source example that includes an entity-context pair from the text (CUI et a. fig. 4 para[0052], generates initial prompt from user query, instructions and content);
.
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the verification method of Cui et al. in order to identify and prevent hallucinations from large language models (Cui et al. para[0002]).
Amoateng et al. does not explicitly disclose providing, as input into the large language model, the source example and a set of user segments to generate alternative examples associated with the source content, each alternative example corresponding to a user segment of the set of user segments.
However Baeuml et al. discloses providing, as input into the large language model, the source example and a set of user segments to generate alternative examples associated with the source content, each alternative example corresponding to a user segment of the set of user segments (Baeuml et al. para[0060], system process NL based input and based on first segment a generates additional LLM output.
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the rendering method of Baeuml et al. in order to reduce latency in generating natural language based output (Baeuml et al. para[0004]).
In regards to claim 17, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the media of claim 16, wherein the particular user segment is identified using a user profile associated with the user interested in the source content (Amoateng et al. para[0027], The user profile may also include information such as topics of interest to the user and other suitable information about the use).
In regards to claim 18, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the media of claim 16, wherein the particular user segment is identified based on a user selection of the particular user segment among user segment options (Amoateng et al. fig. 4 para[0076], responsive to the selection of one of the selectable pills, causing the response to the pill prompt that corresponds to the selection to be displayed to the first user.).
In regards to claim 19, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the media of claim 16, wherein the alternative example is presented as a supplement to the source content (Amoateng et al. para[0058], In some examples, when the user selects a pill, the response to the information request is provided as text or other information in a pop-up window).
In regards to claim 20, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the media of claim 16, wherein the alternative example is presented by modifying the source content to include the alternative example (Amoateng et al. para[0058], When the user selects this global pill prompt, the display of the personalized content feed is changed so that only those feed items that correspond to important things that the user has missed this week are displayed in the feed.).
Claim(s) 2-4, 9, and 13 is/are rejected under 35 U.S.C. 103 as being unpatentable over Amoateng et al. in view of Cui et al., Baeuml et al. and Laban et al. ("SummaC: Revisiting NLI-based models for Inconsistency Detection in Summarization") as made of reference in IDS dated 1/24/2024.
In regards to claim 2, Amoateng et al. as modified by Cui and Baeuml et al. discloses the computing system of claim 1.
Amoateng et al does not explicitly disclose wherein the operations further comprise performing verification of the alternative example by:
obtaining search results associated with the alternative example; and
determining that a threshold number of the search results match at least a portion of the alternative example.
However Laban et al. discloses wherein the operations further comprise performing verification of the alternative example by:
obtaining search results associated with the alternative example (Laban et al. pg164 section2.3 para2, QAG methods follow three steps: (1) question generation (QG), (2) question answering (QA) with the document and the summary, (3) matching document and summary answers. ); and
determining that a threshold number of the search results match at least a portion of the alternative example (Laban et al. pg164 section 2.3 para2, A summary is considered consistent if few or no questions have differing answer with the document.).
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the verification method of Laban et al. in order to identify correct factual errors in personalized summaries (Laban et al. pg163 section1 para2).
In regards to claim 3, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the computing system of claim 1, wherein the operations further comprise performing verification of the alternative example by:
obtaining search results, from a search engine, identified as relevant to a query, the query including at least a portion of the alternative example (Amoateng et al. para[0063], For example, the personalized feed or a subset of the personalized feed may be provided to the user while doing searches, while using communication or social networking applications, while using various applications, and during other suitable user experiences that are associated with the service);
segmenting reference documents associated with the search results into portions (Amoateng et al. para[0065], By providing a personalized feed via feed services, including a personalized title and summary, and bundling feed items in a personalized manner, the cognitive load of each user is reduced, the user's time is saved, the user receives more relevant and useful results, meaningful groupings of feed items are provided, and the feed allows for interactive consumption of content rather than random scrolling);
removing portions of the reference documents identified as semantically dissimilar to the at least the portion of the alternative example (Amoateng et al. para[0055], For example, the post processing of the output of the function call to the large language model may include parsing, removing words, characters and the like that will not be included and the feed, and cleaning up the output by adding text that will be used to fill out the output before the text is shown in the user's feed).
Amoateng et al does not explicitly disclose for each reference document, determining if at least one portion of the reference document matches the at least the portion of the alternative example; and
identifying that a threshold number of the reference documents are determined to have the at least one portion of the reference document matching the at least the portion of the alternative example.
However Laban et al. discloses for each reference document, determining if at least one portion of the reference document matches the at least the portion of the alternative example (Laban et al. pg164 section2.3 para2, QAG methods follow three steps: (1) question generation (QG), (2) question answering (QA) with the document and the summary, (3) matching document and summary answers. ); and
identifying that a threshold number of the reference documents are determined to have the at least one portion of the reference document matching the at least the portion of the alternative example (Laban et al. pg164 section 2.3 para2, A summary is considered consistent if few or no questions have differing answer with the document).
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the verification method of Laban et al. in order to identify correct factual errors in personalized summaries (Laban et al. pg163 section1 para2).
In regards to claim 4, Amoateng et al. as modified by Cui et al., Baeuml et al. and Laban et al. The computing system of claim 2, wherein the verification of the alternative example further comprises obtaining an input indicating the alternative example is accurate prior to providing the alternative example for display (Laban et al. pg164 section 2.1 para1, the model must then retrieve relevant evidence and decide if whether the claim is supported, refuted or if there is not enough information in the corpus).
In regards to claim 9, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the computing system of claim 1.
Amoateng et al does not explicitly disclose wherein the alternative example comprises an alternative entity different from the entity of the source example and a claim generated based on the context of the source example.
However Laban et al. discloses wherein the alternative example comprises an alternative entity different from the entity of the source example and a claim generated based on the context of the source example (Laban et al. pg165 section3.1 para3, each premise-hypothesis combination is run through the NLI model, which produces a probability distribution over the three NLI categories for entailment, contradiction, and neutral).
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the verification method of Laban et al. in order to identify correct factual errors in personalized summaries (Laban et al. pg163 section1 para2).
In regards to claim 13, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the method of claim 11,
Amoateng et al. does not explicitly disclose wherein the alternative example includes an alternative entity different from the entity of the source content and a claim that matches the context of the source content.
However Laban et al. discloses wherein the alternative example includes an alternative entity different from the entity of the source content and a claim that matches the context of the source content (Laban et al. pg165 section3.1 para3, each premise-hypothesis combination is run through the NLI model, which produces a probability distribution over the three NLI categories for entailment, contradiction, and neutral).
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the verification method of Laban et al. in order to identify correct factual errors in personalized summaries (Laban et al. pg163 section1 para2).
Claim(s) 5, 14-15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Amoateng et al. in view of Cui et al., Baeuml et al., and Fernandez-Ruiz (US2014/0372415).
In regards to claim 5, Amoateng et al. as modified by Cui et al. and Baeuml et al. discloses the computing system of claim 1.
Amoateng et al. does not explicitly disclose wherein the operations further comprise obtaining user feedback associated with the alternative example, wherein the user feedback is provided in response to a display of the alternative example and indicates an approval or a disapproval of the alternative example.
However Fernandez-Ruiz discloses wherein the operations further comprise obtaining user feedback associated with the alternative example, wherein the user feedback is provided in response to a display of the alternative example and indicates an approval or a disapproval of the alternative example (Fernandez-Ruiz para[0041], When a user is presented with requested content and the additional content, observations of the user's course of actions with respect to the additional content may be used to better-tailor selections and presentation of subsequent additional content).
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the content enrichment method of Fernandez-Ruiz in order to locate and integrate additional content of interest to the user (Fernandez-Ruiz para[0039]).
In regards to claim 14, Amoateng et al. as modified by Cui et al. discloses the method of claim 11.
Amoateng et al. does not explicitly disclose further comprising receiving user feedback associated with the alternative example, wherein the user feedback is provided in response to a display of the alternative example and indicates an approval or a disapproval of the alternative example.
However Fernandez-Ruiz discloses further comprising receiving user feedback associated with the alternative example, wherein the user feedback is provided in response to a display of the alternative example and indicates an approval or a disapproval of the alternative example (Fernandez-Ruiz para[0041], When a user is presented with requested content and the additional content, observations of the user's course of actions with respect to the additional content may be used to better-tailor selections and presentation of subsequent additional content).
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the content enrichment method of Fernandez-Ruiz in order to locate and integrate additional content of interest to the user (Fernandez-Ruiz para[0039]).
In regards to claim 15, Amoateng et al. as modified by Cui et al. and Fernandez-Ruiz discloses the method of claim 14, wherein the user feedback is used to supplement or modify interests included in a user profile (Fernandez-Ruiz para[0041], This feedback enables the system, for example, to learn the interests of the user).
It would have been obvious to one of ordinary skill in the art before the filing date of the invention to have combined the personalization method of Amoateng et al. with the content enrichment method of Fernandez-Ruiz in order to locate and integrate additional content of interest to the user (Fernandez-Ruiz para[0039]).
Response to Arguments
Applicant’s arguments, see pg2, filed 02/23/2026, with respect to Claims 1-15 have been fully considered and are persuasive. The 101 rejection of 11/21/2025 has been withdrawn.
Applicant’s arguments with respect to claims 1-20 have been considered but are moot because the arguments do not apply the current rejection.
Conclusion
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 NICHOLAS HASTY whose telephone number is (571)270-7775. The examiner can normally be reached Monday-Friday 8:30am-5:00pm.
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, Matt Ell can be reached at (571)270-3264. 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.
/N.H/Examiner, Art Unit 2141
/MATTHEW ELL/Supervisory Patent Examiner, Art Unit 2141
Prosecution Insights