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 office action is in responsive to communication(s):
Application filed on 5/7/2024 with effective filing date of 5/7/2024.
The status of the claims is summarized as below:
Claims 1-10, 23-32 are pending (based on preliminary amendment filed on 7/16/2024).
Claims 1, and 23 are independent claims.
Information Disclosure Statement
The information disclosure statement(s) filed on 7/16/2024, 5/26/2026 comply/complies with the provisions of 37 C.F.R. § 1.97, 1.98, and MPEP § 609, and therefore has/have been placed in the application file. The information referred to therein has/have been considered as to the merits.
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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) 1-3, 6-7, 9, 23-25, 28-29, 31 is/are rejected under 35 U.S.C. 103 as being unpatentable over Dicklin et al. (US Pub 20250103640, from IDS filed 5/26/2026, hereinafter Dicklin), in view of Bax et al. (US Pub 20250005266, hereinafter Bax).
Per claim 1, Dicklin teaches:
A method of determining a citation for output content generated by a trained generative machine learning model (ML model), the method comprising: (abstract: providing generative answers with citations to user query based on documents in cloud storage);
determining a first set of one or more content items by filtering content items in a content database based on a text prompt embedding, wherein the text prompt embedding is generated based on a text prompt used by the ML model to generate the output content; ([0117-0120] Fig. 6 shows at step 650 a subset of documents in the cloud storage is filtered based on query embedding generated at step 640 (text prompt embedding), where the query embedding was generated based on a generative MLM prompt (a text prompt) obtained from the user interface in step 610 and the response (output content) from step 630);
identifying a base content item for the output content by filtering the ([0117-0120] Fig. 6 shows at step 650, portions of the subset of documents are identified based on a threshold similarity (component feature), where the threshold similarity is generated based on a comparison between the query embedding of the response (output content) from 640 and the query embedding of the portion of each document portion; [0113-0114] the response would highlight portion/region to help user to determine parts of the response supported by the document in the subset; also see [0111-0113] Fig. 5);
transmitting a citation for the output content based on the identified base content item. ([0120] Fig. 6 shows at step 660, one or more citations to the identified document portions are included in the output; also see [0111-0113] Fig. 5).
Although Dicklin teaches filtering for source documents once using prompt/query embedding, Dicklin does not explicitly teach “determining a second set of one or more content items by filtering the first set of content items using an output content embedding, wherein the output content embedding is generated based on the output content”.
However, Bax teaches:
determining a second set of one or more content items by filtering the first set of content items using an output content embedding, wherein the output content embedding is generated based on the output content; ([0035-0036] Fig. 1 shows a text processor 110 including a statement identifier 112 and citation generator 114, where a top subset of sources (a second set of content items) is determined by filtering/ranking a list of possible sources using the statement identifier 112 (output content embedding) to identify statements from the generated content (output content), and the citation generator 114 then use each statement as a query to data sources to receive candidate sources, and score/rank each sources based on its relevancy to each statement).
Bax and Dicklin are analogous art because Bax also teaches method of including automated citations for generated content. Therefore, it would have been obvious to one of ordinary skills in art before the effective filing date, having the teachings of Dicklin and Bax before him/her, to modify the teachings of Dicklin to include the teachings of Bax so that the first filtered set of document using prompt embedding can be further filtered using statement from the generated content. One would be motivated to make the combination, with a reasonable expectation of success, because it would provide additional filtering for finding citations based on the generated content, which would further improve accuracy for the automatic generated citations.
Per claim 2, Dicklin-Bax teach all the limitations from claim 1, and further teach:
wherein the output content comprises an image generated by the ML model. (Bax [0043] the output of the LLM may comprise other media type such as image).
Per claim 3, Dicklin-Bax teach all the limitations from claim 1, and they further teach:
wherein the content items of the content database are clustered by content style; (Bax [0053-0054] from step 210 of Fig. 2, sources can include web documents that are clustered/grouped based on content type/style such as text, images, video, audio);
wherein the determining the second set of content items comprises filtering the first set of content items based on the clustered content items and using a style embedding generated based on the output content and a component embedding generated based on the output content. (Bax [0054-0058] filtering of content information can be based on web pages clustered/grouped by content type/style, with trustworthiness scores assigned to each of the results from statement identified from the output, and relevancy score for each result which are then ranked; also see Fig. 4 [0078-0087]).
Per claim 6, Dicklin-Bax teach all the limitations from claim 1, and further teach:
wherein the citation comprises attribution information indicating a title, a creator, or a source of the base content item. (Bax [0009] sources of a generated text may be used to annotate the generated text with citations).
Per claim 7, Dicklin-Bax teach all the limitations from claim 1, and further teach:
wherein the citation comprises a similarity score indicating a degree of similarity between the output content and the base content item. (Bax [0083-0087] scores of the results based on their relevancy to the statements from statement identifier can be included in the search result of each statement).
Per claim 9, Dicklin-Bax teach all the limitations from claim 1, and they further teach:
wherein the citation comprises a human-perceptible watermark provided on the output content indicating the base content or indicating ownership of the base content. ([0132-0133] Fig. 8 shows the generated content can includes citation text content 808 (human perceptible watermark), that indicate the base content such as which documents support which text content).
Per claim 23, claim 23 is a system claim with memory (Dicklin Fig. 14, memory 1404, 1406, and 1416) and control circuitry (Dicklin Fig. 14, control circuitry/ processing device 1402), that includes a method substantially the same as the method from claim 1, which are likewise rejected as claim 1; and an additional limitations of “… and to store the second set of one or more content items in the memory …” (taught by Bax [0065-0068]: step 220 from Fig. 2 shows results are stored and displayed based on user selections).
Per claims 24-25, 28-29, 31, claims 24-25, 28-29 and 31 include limitations that are substantially the same as claim 2-3, 6-7, 9 respectively, and are likewise rejected.
Claim(s) 4-5, 26-27 is/are rejected under 35 U.S.C. 103 as being unpatentable over Dicklin, in view of Bax, and Kennedy et al. (US Pub 20120136853, hereinafter Kennedy).
Per claim 4, Dicklin-Bax teach all the limitations from claim 1, but they do not teach “wherein the identifying the base content item further comprises: matching, using a scale-invariant feature transform (SIFT) process, the base content item and the output content; and obtaining matched SIFT locations in the output content to generate a mask of matched features”.
However, Kennedy teaches:
wherein the identifying the base content item further comprises:
matching, using a scale-invariant feature transform (SIFT) process, the base content item and the output content; and ([0094] SIFT is used to extract feature from images; given two images, each with a set of SIFT interest points,, the invention may be able to identify match points between the two images);
obtaining matched SIFT locations in the output content to generate a mask of matched features. ([0094] the location/interest point of the matches may be identified and obtained).
Kennedy and Dicklin-Bax are analogous art because Kennedy also teaches method of managing content with citations. Therefore, it would have been obvious to one of ordinary skills in art before the effective filing date, having the teachings of Dicklin-Bax and Kennedy before him/her, to modify the teachings of Dicklin-Bax to include the teachings of Kennedy so that the citation information can be obtained for matched images. One would be motivated to make the combination, with a reasonable expectation of success, because it would provide citations for images for machine generative models.
Per claim 5, Dicklin-Bax-Kennedy teach all the limitations from claim 4, and further teach:
wherein based on the obtained matched SIFT locations, the citation indicates a first portion of the output content of greater similarity to the base content item than a second portion of the output content item. (Kennedy [0094] SIFT is used to compare images and obtain matched interest points on the images; Bax: [0035-0036] a list of ranked citation based on similarity are shown to the user based on the relevancy of the statements in the output compared to the sources).
Per claims 26-27, claims 26-27 include limitations that are substantially the same as claim 4-5 respectively and are likewise rejected.
Claim(s) 8, 10, 30, 32 is/are rejected under 35 U.S.C. 103 as being unpatentable over Dicklin, in view of Bax, and Tehranchi et al. (US Pub 20120136853, hereinafter Tehranchi).
Per claim 8, Dicklin-Bax teach all the limitations from claim 1, but they do not teach “wherein the citation comprises copyright restriction information for the base content item“.
However, Tehranchi teaches:
wherein the citation comprises copyright restriction information for the base content item ([0209] citation capability may be implemented with connection to external source, including carrying a “copywrite” flag).
Tehranchi and Dicklin-Bax are analogous art because Tehranchi also teaches method of managing content with citations. Therefore, it would have been obvious to one of ordinary skills in art before the effective filing date, having the teachings of Dicklin-Bax and Tehranchi before him/her, to modify the teachings of Dicklin-Bax to include the teachings of Tehranchi so that the citation information can be configured to include copyright information. One would be motivated to make the combination, with a reasonable expectation of success, because it would provide configurable citation format to include the necessary information such as copyright restriction.
Per claim 10, Dicklin-Bax teach all the limitations from claim 1, but they do not teach “wherein the citation comprises a machine-detectable watermark imperceptible to humans using a naked eye, and provided on the output content indicating the base content or indicating a source of the base content“.
However, Tehranchi teaches:
wherein the citation comprises a machine-detectable watermark imperceptible to humans using a naked eye, and provided on the output content indicating the base content or indicating a source of the base content. ([0209] electronic citation can include watermark payload that include the locations of copyrighted material as part of the signal of a media content).
Tehranchi and Dicklin-Bax are analogous art because Tehranchi also teaches method of managing generative contents. Therefore, it would have been obvious to one of ordinary skills in art before the effective filing date, having the teachings of Dicklin-Bax and Tehranchi before him/her, to modify the teachings of Dicklin-Bax to include the teachings of Tehranchi so that the citation information/format can be configured to include machine detectable watermark. One would be motivated to make the combination, with a reasonable expectation of success, because it would provide configurable citation format to include hidden/watermark information to denote “copyright” material and its locations, and also help to provide a more detailed citation information in the watermark payload (Tehranchi [0209]).
Per claims 30, 32, claims 30 and 32 include limitations that are substantially the same as claim 8 and 10 respectively, and are likewise rejected.
Conclusion
The examiner requests, in response to this Office action, support by shown for language added to any original claims on amendment and any new claims. That is, indicate support for newly added claim language by specifically pointing to page(s) and line no(s) in the specification and/or drawing figure(s). This will assist the examiner in prosecuting the application.
When responding to this office action, Applicant is advised to clearly point out the patentable novelty which he or she thinks the claims present, in view of the state of the art disclosed by the references cited or the objections made. He or she must also show how the amendments avoid such references or objections, See 37 CFR 1.111(c).
Contact Information
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PHOEBE X PAN whose telephone number is (571)270-7794. The examiner can normally be reached M-F 9am-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, Fred Ehichioya can be reached on (571) 272-4034. 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.
/PHOEBE X PAN/Examiner, Art Unit 2179
/IRETE F EHICHIOYA/Supervisory Patent Examiner, Art Unit 2179
Prosecution Insights