DETAILED ACTION
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 .
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 15-20 rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more. Using the subject matter eligibility test from page 74621 of the Federal Register Notice titled “2014 Interim Guidance on Patent Subject Matter Eligibility,” a two-step process is performed. Under step 1, the claims are analyzed to determine if the claim is directed to a process, machine, article of manufacture, or composition of matter. In this case, claims 15-20 are directed to a method, which is a process. Step 2A (part 1 of the Mayo test), using the guidance from pages 50-57 of the Federal Register Vol. 84 No. 4 from Monday, January 7, 2019, requires applying a two-prong inquiry. In Prong One, examiners evaluate whether the claim recites a judicial exception, determining if the claim is directed to a law of nature, a natural phenomenon, or an abstract idea. In this case, claims 15 recites predicting a document, which is a mental process. In Prong Two, examiners evaluate whether the judicial exception is integrated into a practical application that imposes a meaningful limit on the judicial exception. In this case, receiving and providing data is mere extrasolution activity, while use of a trained model is simply applying a computer to perform the mental process, neither of which incorporates the mental process into a practical application.
Step 2B (part 2 of the Mayo test) requires analyzing the claims to determine if they recite additional elements that amount to significantly more than the judicial exception. In this case, the claims do not include additional elements that are sufficient to amount to significantly more than the abstract idea itself.
Regarding claim 15, predicting a document is a mental process, which is an abstract idea. Additional limitations of receiving and providing data is mere extrasolution activity, while use of a trained model is simply applying a computer to perform the mental process, neither of which incorporates the mental process into a practical application or constitutes significantly more.
Regarding claims 16-19, the limitations are further clarifications of the above abstract ideas.
Regarding claim 20, determining a request is a mental process, which is an abstract idea. Additional limitations of sending, receiving, and providing the request are mere extrasolution activity, and do not integrate the abstract idea into a practical application or constitute significantly more.
The limitations of the claims, taken alone, do not amount to significantly more than the above-identified judicial exception (the abstract idea). Looking at the limitations as an ordered combination adds nothing that is not already present when looking at the elements individually. Applicable case law cited in the Federal Register includes, but is not limited to: Alice Corp., 134 S. Ct. at 2355-56, Digitech Image Tech., LLC v. Electronics for Imaging, Inc., 758 F.3d 1344 (Fed. Cir. 2014), Benson, 409 U.S. at 63.
See "Preliminary Examination Instructions in view of the Supreme Court Decision in Alice Corporation Pty. Ltd. v. CLS Bank International, et al.," dated June 25, 2014, and the Federal Register notice titled "2014 Interim Guidance on Patent Subject Matter Eligibility" (79 FR 74618).
Claim Rejections - 35 USC § 102
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.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claim(s) 1-2, 5-6, 10-11, 14-15, 18, and 21 is/are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Bonifacio et al. (Bonifacio, L., Abonizio, H., Fadaee, M., & Nogueira, R. (2022). Inpars: Data augmentation for information retrieval using large language models. arXiv preprint arXiv:2202.05144.), hereinafter referred to as Bonifacio.
Regarding claim 1, Bonifacio teaches:
A computer-implemented method for prompt-based query generation, comprising:
receiving, by a computing device, at least two prompts associated with a retrieval task to be performed on a corpus of documents associated with the task (page 4 first two full paragraphs, where two prompt templates are used for generating questions, and page 3 section 4, where data retrieval is trained on datasets);
applying, based on the at least two prompts and the corpus of documents, a large language model to generate a synthetic training dataset comprising a plurality of query-document pairs, wherein each query-document pair comprises a synthetically generated query and a document from the corpus of documents (pages 4-5,last three paragraphs of section 4.2, where a GPT-3 model is used to generate questions using the two prompts and the datasets, to generate 100,000 questions with their respective input documents);
training, on the plurality of query-document pairs from the synthetic training dataset, a document retrieval model to take an input query associated with the retrieval task and predict an output document retrieved from the corpus of documents (page 3 section 4, where data retrieval is trained on the synthetic data, and section 4.3 first three paragraphs, where the retrieval model is trained and finetuned using the synthetic questions); and
providing, by the computing device, the trained document retrieval model (page 8 section 7 first full paragraph, where the method is presented).
Regarding claim 2, Bonifacio teaches:
The computer-implemented method of claim 1, further comprising:
determining, by the document retrieval model and for each query-document pair, a retrieval score, wherein the retrieval score is indicative of a relevance of the document of the query- document pair to the query of the query-document pair (page 5 second column, second full paragraph, where a relevance probability for each passage is identified, interpreted as the retrieval score).
Regarding claim 5, Bonifacio teaches:
The computer-implemented method of claim 1, wherein the plurality of query-document pairs comprise noisy data (page 8 section 6.3, where lower probability pairs are considered noisy data), and further comprising:
filtering the plurality of query-document pairs to remove the noisy data (page 8 section 6.3, where the pairs are filtered down to 10,000).
Regarding claim 6, Bonifacio teaches:
The computer-implemented method of claim 5, further comprising:
fine-tuning the document retrieval model based on the plurality of filtered query-document pairs (page 5 first full paragraph, where the filtered pairs are used for finetuning the models).
Regarding claim 10, Bonifacio teaches:
The computer-implemented method of claim 1, wherein for each document in the corpus, a query is sampled based on a distribution comprising a tunable temperature hyperparameter (pages 4-5, paragraph spanning pages, where documents are sampled with a temperature parameter of 0), and further comprising:
maintaining the temperature hyperparameter below a temperature threshold to generate a diverse array of queries (pages 4-5, paragraph spanning pages, where a temperature of 0 is used which defaults to greedy decoding).
Regarding claim 11, Bonifacio teaches:
The computer-implemented method of claim 1, wherein the retrieval task comprises one or more of: question-to-document retrieval, a question-to-question retrieval, a claim-to-document retrieval, an argument-support-document retrieval, or a counter-argument- support-document retrieval (pages 2-3 paragraph spanning pages, where the retrieval models are finetuned using question-document pairs, and page 5 second full paragraph, where query document pairs are used).
Regarding claim 14, Bonifacio teaches:
The computer-implemented method of claim 1, wherein the at least two prompts is less than eight prompts (page 4 first two full paragraphs, where two prompt templates are used for generating questions).
Regarding claim 15, Bonifacio teaches:
A computer-implemented method of applying a trained document retrieval model, comprising:
receiving, by a computing device, an input query associated with a retrieval task to be performed on a corpus of documents associated with the task (page 5 section 4.3 second paragraph, where each query results in 1000 candidate documents being retrieved);
predicting, by the trained document retrieval model, a document from the corpus of documents, wherein the document has a high relevance to the input query (page 5 second column, second full paragraph, where a relevance probability for each passage is identified, interpreted as the retrieval score), the document retrieval model having been trained on a plurality of query-document pairs from a synthetic training dataset (page 3 section 4, where data retrieval is trained on the synthetic data, and section 4.3 first three paragraphs, where the retrieval model is trained and finetuned using the synthetic questions), the synthetic training dataset having been generated by a large language model based on at least two prompts associated with the retrieval task (pages 4-5,last three paragraphs of section 4.2, where a GPT-3 model is used to generate questions using the two prompts and the datasets, to generate 100,000 questions with their respective input documents, and page 4 first two full paragraphs, where two prompt templates are used for generating questions); and
providing, by the computing device and in response to the input query, the predicted document (page 5 section 4.3 second paragraph, where each query results in 1000 candidate documents being retrieved).
Regarding claim 18, Bonifacio teaches:
The computer-implemented method of claim 15, wherein the retrieval task comprises one or more of: question-to-document retrieval, a question-to-question retrieval, a claim-to-document retrieval, an argument-support-document retrieval, or a counter-argument- support-document retrieval (pages 2-3 paragraph spanning pages, where the retrieval models are finetuned using question-document pairs, and page 5 second full paragraph, where query document pairs are used).
Regarding claim 21, Bonifacio teaches:
A computing device, comprising:
one or more processors (page 1 section 1 first two paragraphs, where computation is performed, requiring a processor); and
data storage, wherein the data storage has stored thereon computer-executable instructions that, when executed by the one or more processors, cause the computing device to carry out functions (page 1 section 1 first two paragraphs, where computation is performed, requiring a memory and instructions) comprising:
receiving, by the computing device, at least two prompts associated with a retrieval task to be performed on a corpus of documents associated with the task (page 4 first two full paragraphs, where two prompt templates are used for generating questions, and page 3 section 4, where data retrieval is trained on datasets);
applying, based on the at least two prompts and the corpus of documents, a large language model to generate a synthetic training dataset comprising a plurality of query-document pairs, wherein each query-document pair comprises a synthetically generated query and a document from the corpus of documents (pages 4-5,last three paragraphs of section 4.2, where a GPT-3 model is used to generate questions using the two prompts and the datasets, to generate 100,000 questions with their respective input documents);
training, on the plurality of query-document pairs from the synthetic training dataset, a document retrieval model to take an input query associated with the retrieval task and predict an output document retrieved from the corpus of documents (page 3 section 4, where data retrieval is trained on the synthetic data, and section 4.3 first three paragraphs, where the retrieval model is trained and finetuned using the synthetic questions); and
providing, by the computing device, the trained document retrieval model (page 8 section 7 first full paragraph, where the method is presented).
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) 3-4, 16, and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Bonifacio, in view of Saxena et al. (US 2022/0382805 A1), hereinafter referred to as Saxena.
Regarding claim 3, Bonifacio teaches:
The computer-implemented method of claim 1,
Bonifacio does not teach:
wherein the document retrieval model is a dual encoder model comprising a query encoder to encode a query and a document encoder to encode a document.
Saxena teaches:
wherein the document retrieval model is a dual encoder model comprising a query encoder to encode a query and a document encoder to encode a document (para [0083], [0119], where an image of a document is embedded into a visual embedding vector and a query is embedded into a semantic embedding vector).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system of Bonifacia by using the embeddings of Saxena (Saxena para [0083]) in the document retrieval of Bonifacia (Bonifacia page 8 section 7 first full paragraph), in order to determine a top y images or documents related to the query (Saxena para [0083]).
Regarding claim 16, Bonifacio teaches:
The computer-implemented method of claim 15,
Bonifacio does not teach:
wherein the document retrieval model is a dual encoder model comprising a query encoder to encode a query and a document encoder to encode a document.
Saxena teaches:
wherein the document retrieval model is a dual encoder model comprising a query encoder to encode a query and a document encoder to encode a document (para [0083], [0119], where an image of a document is embedded into a visual embedding vector and a query is embedded into a semantic embedding vector).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system of Bonifacia by using the embeddings of Saxena (Saxena para [0083]) in the document retrieval of Bonifacia (Bonifacia page 8 section 7 first full paragraph), in order to determine a top y images or documents related to the query (Saxena para [0083]).
Regarding claim 19, Bonifacio teaches:
The computer-implemented method of claim 15,
Bonifacio does not teach:
wherein the input query is a voice input, and wherein the retrieval task is performed by a computer-implemented intelligent voice assistant.
Saxena teaches:
wherein the input query is a voice input, and wherein the retrieval task is performed by a computer-implemented intelligent voice assistant (Fig. 13, para [0133], where a voice query is input to a voice assistant to perform information retrieval).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system of Bonifacia by using the voice interaction of Saxena (Saxena para [0133]) in the document retrieval of Bonifacia (Bonifacia page 8 section 7 first full paragraph), in order to perform information retrieval via a voice assistant (Saxena para [0133]).
Claim(s) 7 is/are rejected under 35 U.S.C. 103 as being unpatentable over Bonifacio, in view of Wang et al. (US 12,271,698 B1), hereinafter referred to as Wang.
Regarding claim 7, Bonifacio teaches:
The computer-implemented method of claim 6, wherein the fine-tuning of the document retrieval model is based on in-batch random negatives (page 5 second full paragraph, where a negative example is randomly selected for finetuning).
Bonifacio does not teach:
a standard softmax loss;
Wang teaches:
a standard softmax loss (col. 23 lines 39-53, where a softmax is used with the negative random samples and cross entropy loss)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system of Bonifacia by using the linkages of Wang (Wang col. 23 lines 23-53)in the finetuning of Bonifacia (Bonifacia page 5 first full paragraph), in order to improve end-to-end accuracy (Wang col. 23 lines 39-53).
Claim(s) 8 is/are rejected under 35 U.S.C. 103 as being unpatentable over Bonifacio, in view of Alberti et al. (Alberti, C., Andor, D., Pitler, E., Devlin, J., & Collins, M. (2019, July). Synthetic QA corpora generation with roundtrip consistency. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (pp. 6168-6173).), hereinafter referred to as Alberti.
Regarding claim 8, Bonifacio teaches:
The computer-implemented method of claim 5,
Bonifacio does not teach:
wherein the filtering comprises one or more of length filtering, prompt filtering, or round-trip filtering.
Alberti teaches:
wherein the filtering comprises one or more of length filtering, prompt filtering, or round-trip filtering (page 6171-6172 paragraph spanning pages, where roundtrip filtering is performed).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system of Bonifacia by using the filtering of Alberti (Alberti para 6171-6172 paragraph spanning pages) on the pairs of Bonifacia (Bonifacia page 8 section 6.3), as the roundtrip filtering is consistently beneficial and outperforms pretraining without filtering (Alberti page 6171-6172 paragraph spanning pages).
Claim(s) 9 and 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Bonifacio, in view of Wei et al. (Wei, J., Bosma, M., Zhao, V. Y., Guu, K., Yu, A. W., Lester, B., ... & Le, Q. V. (2021). Finetuned language models are zero-shot learners. arXiv preprint arXiv:2109.01652.), hereinafter referred to as Wei.
Regarding claim 9, Bonifacio teaches:
The computer-implemented method of claim 1,
Bonifacio does not teach:
wherein the large language model is a fine-tuned language net (FLAN).
Wei teaches:
wherein the large language model is a fine-tuned language net (FLAN) (page 2 second paragraph, where a finetuned language net is used).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system of Bonifacia by using the model of Wei (Wei page 2 second paragraph) as the large language model of Bonifacia (Bonifacia pages 4-5,last three paragraphs of section 4.2), in order to improve the zero-shot performance of LLMs and expand their reach to a broader audience (Wei page 2 second paragraph)
Regarding claim 17, Bonifacio teaches:
The computer-implemented method of claim 15,
Bonifacio does not teach:
wherein the large language model is a fine-tuned language net (FLAN).
Wei teaches:
wherein the large language model is a fine-tuned language net (FLAN) (page 2 second paragraph, where a finetuned language net is used).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system of Bonifacia by using the model of Wei (Wei page 2 second paragraph) as the large language model of Bonifacia (Bonifacia pages 4-5,last three paragraphs of section 4.2), in order to improve the zero-shot performance of LLMs and expand their reach to a broader audience (Wei page 2 second paragraph)
Claim(s) 12 is/are rejected under 35 U.S.C. 103 as being unpatentable over Bonifacio, in view of Mass et al. (US 2021/0390418 A1), hereinafter referred to as Mass.
Regarding claim 12, Bonifacio teaches:
The computer-implemented method of claim 1, further comprising:
Bonifacio does not teach:
pre-training the document retrieval model on unsupervised general-domain data.
Mass teaches:
pre-training the document retrieval model on unsupervised general-domain data (para [0019], where the BERT document retrieval model is pretrained on a large dataset before finetuning on more specific datasets, where unsupervised techniques are used).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system of Bonifacia by using the pretraining of Mass (Mass para [0019]) on the document retrieval model of Bonifacia (Bonifacia page 8 section 7 first full paragraph), in order to perform reranking for greater accuracy (Mass para [0019]).
Claim(s) 13 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Bonifacio, in view of Morlen et al. (US 9,710,615 B1), hereinafter referred to as Morlen.
Regarding claim 13, Bonifacio teaches:
The computer-implemented method of claim 1,
providing the large language model and the document retrieval model to the organization (page 8 section 7 first full paragraph, where the method is presented), and
Bonifacio does not teach:
wherein the retrieval task and the corpus of documents are maintained behind a firewall of an organization, and further comprising:
wherein the receiving, applying, and training are performed behind the firewall of the organization.
Morlen teaches:
wherein the retrieval task and the corpus of documents are maintained behind a firewall of an organization (Fig. 9, col. 11 line 48 - col. 12 line 12, where document access and retrieval are performed behind a firewall), and further comprising:
wherein the receiving, applying, and training are performed behind the firewall of the organization (Fig. 9, col. 11 line 48 - col. 12 line 12, where document access and retrieval are performed behind a firewall).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system of Bonifacia by using the firewall of Morlen (Morlen col. 11 line 48 - col. 12 line 12) in the document retrieval model of Bonifacia (Bonifacia page 8 section 7 first full paragraph), in order to protect resources of a private network by enforcing security policies (Morlen col. 11 line 48 - col. 12 line 12).
Regarding claim 20, Bonifacio teaches:
The computer-implemented method of claim 15, further comprising:
Bonifacio does not teach:
determining, by the computing device, a request to respond to the input query;
sending the request from the computing device to a second computing device, the second computing device comprising a trained version of the document retrieval model;
after sending the request, the computing device receiving, from the second computing device, the predicted document, and
wherein the providing of the predicted document comprises providing the predicted document as received from the second computing device.
Morlen teaches:
determining, by the computing device, a request to respond to the input query (col. 10 lines 6-33, where a request is received from a user);
sending the request from the computing device to a second computing device, the second computing device comprising a trained version of the document retrieval model (col. 10 lines 6-33, where the request is sent to access a document);
after sending the request, the computing device receiving, from the second computing device, the predicted document (col. 10 lines 6-33, where access to the document is granted), and
wherein the providing of the predicted document comprises providing the predicted document as received from the second computing device (col. 10 lines 6-33, where access to the document is granted).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system of Bonifacia by using the access of Morlen (Morlen col. 10 lines 6-33) in the document retrieval model of Bonifacia (Bonifacia page 8 section 7 first full paragraph), in order to allow an owner of a secure online repository to grant access rights to users (Morlen col. 10 lines 6-33).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Bonifacio et al. (Bonifacio, L., Abonizio, H., Fadaee, M., & Nogueira, R. (2022, July). Inpars: Unsupervised dataset generation for information retrieval. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (pp. 2387-2392).) Abstract teaches harnessing few-shot capabilities of large pretrained language models as synthetic data generators for information retrieval tasks.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BRYAN S BLANKENAGEL whose telephone number is (571)270-0685. The examiner can normally be reached 8:00am-5:30pm.
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, Richemond Dorvil can be reached at 571-272-7602. 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.
/BRYAN S BLANKENAGEL/Primary Examiner, Art Unit 2653
