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
Claims 1-20 are pending. Claims 1 and 11 are independent.
This Application was published as US 20250284721.
Apparent priority is 8 March 2024.
The instant Application is directed to preprocessing information with a machine learning model and then inputting it to a retrieval augmented generation system.
Information Disclosure Statement
The information disclosure statement filed 1/10/2025 fails to comply with 37 CFR 1.98(a)(3)(i) because it does not include a concise explanation of the relevance, as it is presently understood by the individual designated in 37 CFR 1.56(c) most knowledgeable about the content of the information, of each reference listed that is not in the English language. It has been placed in the application file, but the foreign references referred to therein have not been considered.
Claim Rejections - 35 USC § 102
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)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 1-2, 9, 11-12, and 19 is/are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Huang et al. (US 20250111159 A1).
Regarding claim 1, Huang discloses: 1. A method comprising: accessing one or more inferences generated using a machine learning (ML) model; ("[005]...receiving a prompt from a user at an endpoint; providing the prompt to an embedding module to generate an embedding for the prompt; …" see also "[0026] ... For example, embeddings model 235 may be a machine learning model that is trained to generate an n-dimensional vector representation of a set of input data..." - the embedding of the prompt reads on an inference generated using a ML model.)
providing an inference input to a retrieval agent of an object store based on the one or more inferences, ("[0005] ...retrieving one or more embeddings from a vector database based on the embedding for the prompt; …" )
wherein: the object store comprises one or more vector stores representing a plurality of reference documents using semantic encodings, and ("[0020] As shown, the embedding database 130 includes embeddings stored in the database that have been generated by an embedding generator 140. … The chunked data is passed to the embedding module 144, which generates embeddings, such as vector representations, for the chunked data. …The embedding module 144 then stores the vector representation in the embedding database 130. ..." )
the retrieval agent performs a similarity search of the one or more vector stores to retrieve a set of passages from the plurality of reference documents based at least in part on similarity of encodings of the inference input and encodings of passages in the plurality of reference documents; ("[0019]... The one or more embeddings are provided to the search module 124, and the search module 124 searches for and retrieves one or more stored embeddings from the embedding database 130 based on the one or more embeddings for the prompt. In various embodiments, the stored embeddings can be selected based on a similarity to the one or more embeddings for the prompt using cosine similarity, semantic search, various similarity algorithms, or other techniques for vector comparison." )
generating a linguistic prompt for a large language model (LLM) having a context including the one or more inferences and the set of passages; ("[0005] ... generating an augmented prompt based on the one or more embeddings and the embedding for the prompt; …" )
applying the LLM to the linguistic prompt to generate a natural language explanation of the one or more inferences; and ("[0005]... generating a response to the augmented prompt using a machine learning model and using a context of the one or more embeddings as a context for the machine learning model; ..." see also "[0022]...In some embodiments, the language model 180 is a generative large language model. ..." )
causing the natural language explanation of the one or more inferences to be stored, ("[0056] ... The object manager 524 may be used to manage objects, such as generating or storing chat histories, associated with or used by the system 500. ..." )
wherein the method is performed by one or more computing devices. ("[0082]... A processor may also be implemented as a combination of computing devices ..." )
Regarding claim 2, Huang discloses: 2. The method of claim 1, further comprising: receiving a natural language query from the user, wherein: the machine learning model generates the one or more inferences based at least in part on a profile of the user, and ("[0045].... In some cases, the prompt and a chat history may be used as inputs into a preprocessing language model to generate a standalone prompt. In certain embodiments, such a standalone prompt may be used in subsequent steps for performing a vector storage database similarity search..." – the standalone prompt is an inference based on the profile of the user )
generating the linguistic prompt comprises adding the natural language query to the linguistic prompt. ("[0005] ... generating an augmented prompt based on the one or more embeddings and the embedding for the prompt; …" )
Regarding claim 9, Huang discloses: 9. The method of claim 1, wherein performing the similarity search comprises: generating a sequence of lexical tokens based on the one or more inferences; ("[0043] Preprocessing may use a local language model or an external model to generate a summary of the relevant contextual information, which may be included in or which may replace the prompt prior to generating an embedding from the prompt that is used for performing searching and retrieval." - preprocessing the query using a local LM also reads on generating an inference; see further: “[0034] The workflow 300 may then proceed to stage 340 where a selection of an embedding model is received. In some cases, a client device may be sent information based on token count, validation of data, and/or other parameters so that a user may view the information and select an embedding model.” Tokenization is a known step of processing text for a ML model.)
generating a search encoding that represents the sequence of lexical tokens; and ("[0019] … The embedding generator 122 receives the prompt from the output module 104 of the client device 110. The embedding generator 122 generates one or more embeddings for the prompt, such as by generating a vector representation of a query, data, and metadata associated with the prompt…" )
selecting, by the retrieval agent, the set of passages based at least in part on similarity of the search encoding and encodings of the set of passages. ("[0019]... The one or more embeddings are provided to the search module 124, and the search module 124 searches for and retrieves one or more stored embeddings from the embedding database 130 based on the one or more embeddings for the prompt. In various embodiments, the stored embeddings can be selected based on a similarity to the one or more embeddings for the prompt using cosine similarity, semantic search, various similarity algorithms, or other techniques for vector comparison." )
Claim 11 is a media claim with limitations corresponding to the limitations of Claim 1 and is rejected under similar rationale. Additionally, “One or more non-transitory computer-readable media storing instructions” of the Claim are taught by Huang (“[0006] Other embodiments provide... non-transitory, computer-readable media comprising instructions that, when executed by one or more processors of a processing system, cause the processing system to perform the aforementioned methods as well as those described herein; ...”)
Claim 12 is a media claim with limitations corresponding to the limitations of Claim 2 and is rejected under similar rationale.
Claim 19 is a media claim with limitations corresponding to the limitations of Claim 9 and is rejected under similar rationale.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 3 and 13 is/are rejected under 35 U.S.C. 103 as being unpatentable over Huang in view of Heinonen (US 20250191031 A1).
Regarding claim 3, Huang discloses: 3. The method of claim 2, wherein: the natural language query comprises a request for a product recommendation, the ("[0042]...For example, whether the prompt relates to a product question…" )
ML model is part of an ML-based recommendation system, (not explicitly disclosed )
the plurality of reference documents comprises a plurality of product descriptions, and (not explicitly disclosed )
the retrieval agent performs the similarity search based at least in part on the natural language query. ("[0045].... In some cases, the prompt and a chat history may be used as inputs into a preprocessing language model to generate a standalone prompt. In certain embodiments, such a standalone prompt may be used in subsequent steps for performing a vector storage database similarity search..." )
Huang does not explicitly disclose that the system is for product recommendation, or that the corpus comprises product descriptions.
Heinonen discloses: ML model is part of an ML-based recommendation system, ("[0159] ... The method enables the recipient to perform optimized targeting, product recommendations, pricing, bundling, and new product introduction based on the optimized expected outcomes 1510, thereby enabling optimization of a business. The method generates combinations of the desired data points for an evaluation system and selects the best message and predicts the expected outcomes 1510. ..." )
and the plurality of reference documents comprises a plurality of product descriptions, and ("[0038] The one or more data sets may include personal data, transaction data, recipient data, marketing data, support data, partner data, and public data. Optionally, the one or more data sets include unstructured text, including product descriptions, recipient history, messaging, etc." )
Huang and Heinonen are considered analogous art to the claimed invention because they disclose RAG systems. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Huang with to recommend products based on product descriptions and a product recommendation request, as taught by Heinonen. Doing so would have been beneficial to improve sales, marketing, and customer service. (Heinonen [0072])
Claim 13 is a media claim with limitations corresponding to the limitations of Claim 3 and is rejected under similar rationale.
Claim(s) 4 and 14 is/are rejected under 35 U.S.C. 103 as being unpatentable over Huang in view of Jiang et al. ("Chameleon: a Heterogeneous and Disaggregated Accelerator System for Retrieval-Augmented Language Models").
Regarding claim 4, Huang does not disclose the additional limitations.
Jiang discloses: 4. The method of claim 1, wherein: the LLM is applied to the linguistic prompt using a query acceleration engine, and the query acceleration engine comprises the retrieval agent and the LLM. ("Chameleon employs heterogeneous hardware to accelerate both LLM inference and vector search efficiently." pg. 4, Section 3)
Huang and Jiang are considered analogous art to the claimed invention because they disclose RAG systems. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Huang the acceleration engine taught by Jiang. Doing so would have been beneficial to improve speed. (Jiang pg. 12, Conclusion)
Claim 14 is a media claim with limitations corresponding to the limitations of Claim 4 and is rejected under similar rationale.
Claim(s) 5-7, 10, 15-17, and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Huang in view of Ananthanarayanan et al. (US 20250274367 A1).
Regarding claim 5, Huang does not disclose the additional limitations.
Ananthanarayanan discloses: 5. The method of claim 1, wherein the ML model is part of an anomaly detection system, ("[0029]...Anomaly detection can be performed using various methodologies including a machine learning model." )
wherein the method further comprises: continuously monitoring, by the anomaly detection system, a series of logs; and ("[0029] The live monitoring can be of various data streams which can be monitored using KQL as an example but can be any query language and any data product. The scheme of the language provides various fields, which include performance fields and granularities at which the fields should be monitored..." see also: "[0016] Using the functionality of the knowledge base assistant, a troubleshooting assistant is disclosed that provides dynamic and real-time network monitoring. The troubleshooting assistant is integrated with live data feeds, including cloud data, system logs, and telemetry data. ..." )
in response to detection of one or more anomalous logs in the series of logs, generating, by the anomaly detection system, a trigger condition; ("[0017] The troubleshooting assistant operates dynamically and proactively, and detects and responds to alerts in real-time…" )
wherein the one or more inferences identify the one or more anomalous logs; and wherein the linguistic prompt is generated in response to the trigger condition. ("[0028] Once the issue is localized, retrieval-augmented generation (RAG) can be used to index information and improve accuracy, using for example tickets, incident reports, sections from troubleshooting data from knowledge base 111 such as troubleshooting guides and manuals, and the like. A prompt 120 for a LLM 121 can be generated. The generation of the prompt can include an index search of the incident reports 110 and knowledge base 111 using metadata and key words from monitoring 101 (e.g., EDRs, performance indicators), detected anomalies and generated alerts, fault localization 103, and user complaints associated with IMSI 104...")
Huang and Ananthanarayanan are considered analogous art to the claimed invention because they disclose RAG systems. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Huang to detect anomalies as taught by Ananthanarayanan. Doing so would have been beneficial to improve network performance and user experience. (Ananthanarayanan [0026])
Regarding claim 6, Huang does not disclose the additional limitations.
Ananthanarayanan discloses: 6. The method of claim 5, wherein: the plurality of reference documents comprises at least one of: manuals, troubleshooting cheat sheets, frequently asked questions (FAQ) documents, discussion forums, tutorials, or weblog posts. ("[0028] Once the issue is localized, retrieval-augmented generation (RAG) can be used to index information and improve accuracy, using for example tickets, incident reports, sections from troubleshooting data from knowledge base 111 such as troubleshooting guides and manuals, and the like..." )
See claim 5 for motivation statement.
Regarding claim 7, Huang discloses: 7. The method of claim 5, wherein performing the similarity search comprises: generating one or more log embeddings based on the one or more anomalous logs; and selecting the set of passages based at least in part on similarity of embeddings of passages in the plurality of reference documents to the one or more log embeddings. ("[0019]... The one or more embeddings are provided to the search module 124, and the search module 124 searches for and retrieves one or more stored embeddings from the embedding database 130 based on the one or more embeddings for the prompt. In various embodiments, the stored embeddings can be selected based on a similarity to the one or more embeddings for the prompt using cosine similarity, semantic search, various similarity algorithms, or other techniques for vector comparison." - it would have been obvious as disclosed in claim 5 to use the system to detect anomalous logs, including generating embeddings as disclosed by Huang.)
Regarding claim 10, Huang does not explicitly disclose an engineered prompt template.
Ananthanarayanan discloses: 10. The method of claim 1, wherein the linguistic prompt is based on an engineered prompt template ("[0025]...In addition to KQLs and metadata, skills can also contain a prompt file. Along with the output of KQLs that are executed at each skill execution frequency, information in the metadata is used to replace the placeholders in the prompt file. The constructed final prompt is run over an LLM to generate network insights around this specific skills capability. A sample prompt file is as follows:…”)
comprising: a natural language command, ("[0025]... META PROMPT:As an LLM, your task is to use the information provided to assess the health of the network. Consider the output of the KQL script execution and use the evaluation criteria to identify any anomalies. If problems are identified, refer to the ′Problem Analysis′ and ′Reference Documents′ sections to understand the possible causes and steps for remediation. Additionally, use the ′Search Result Analysis′ to get further insights into the problem. Your ultimate aim is to ensure the optimal health and performance of the network…”)
one or more guardrails limiting a scope of output, ("[0025]... // Grounding Data from Search Results
SEARCH RESULT ANALYSIS: Based on the search results, the following additional insights were gathered: <<GROUNDING DATA FROM SEARCH RESULTS>>…”)
a context portion, and ("[0025]... REFERENCE DOCUMENTS: For further understanding and troubleshooting steps, refer to the following documents: <<REFERENCES>>…”)
a user query portion. ("[0025]... PROBLEM ANALYSIS: Based on the anomalies identified in the network performance, the potential issues could be due to <<SOLUTION_DESCRIPTION>>…”)
Huang and Ananthanarayanan are considered analogous art to the claimed invention because they disclose RAG systems. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Huang to use engineered prompts as taught by Ananthanarayanan. Doing so would have been beneficial to offer a comprehensive and proactive approach. (Ananthanarayanan [0026])
Claim 15 is a media claim with limitations corresponding to the limitations of Claim 5 and is rejected under similar rationale.
Claim 16 is a media claim with limitations corresponding to the limitations of Claim 6 and is rejected under similar rationale.
Claim 17 is a media claim with limitations corresponding to the limitations of Claim 7 and is rejected under similar rationale.
Claim 20 is a media claim with limitations corresponding to the limitations of Claim 10 and is rejected under similar rationale.
Claim(s) 8 and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Huang in view of Knutsson et al. (US 20200234305 A1).
Regarding claim 8, Huang does not explicitly disclose the additional limitations.
Knutsson discloses: 8. The method of claim 1, wherein the ML model is part of a fraud detection system, ("[0001] The invention pertains to the technical field of systems for detecting fraud in financial transactions." )
wherein the method further comprises: continuously monitoring, by the fraud detection system, a series of financial transactions; and ("[0105] FIG. 1 illustrates an example workflow relating to the present invention, with reference signs (1)-(7) referred to accordingly below. Data on current transactions (1) is fed to the system. ..." - see also "[0002]... Such systems may operate in either batch (processing transactions as a group of files at periodic times during the day) or real time mode (processing transactions one at a time, as they enter the system)…" )
in response to detection of one or more anomalous transactions in the series of financial transactions, generating, by the fraud detection system, a trigger condition; ("[0105]...The system further comprises a second model module (3) which reprioritizes the list of alerted transactions…" -alerted transactions reads on a trigger )
wherein the one or more inferences comprise the one or more anomalous transactions; and ("[0105]...Hereby, note that the second model module (3) may use the output of the first model module (2) and thus work together with the first model in a dependent fashion…" )
wherein the linguistic prompt is generated in response to the trigger condition. ("[0106] Hereby, the second model concerns a machine learning model, e.g. gradient boosting machine, random forest or support vector machine..." see also: "[0014]... Hereby, the term “machine learning algorithm”, in its broadest interpretation, refers to any data processing algorithm capable of processing the output of the at least one detection method in a fashion that enables taking into account correlation between the output of at least a first and a second detection method belonging to said at least one detection method, preferably at least two detection methods..." – Huang discloses the linguistic prompt for a machine learning model, and Knutsson discloses that the machine learning model is invoked in response to the trigger (see fig. 1))
Huang and Knutsson are considered analogous art to the claimed invention because they disclose machine learning systems. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Huang to use the RAG system taught by Huang as a second stage model to detect financial fraud in the system taught by Knutsson. Doing so would have been beneficial to decrease the mental and physical effort of the user. (Knutsson [0015])
Claim 18 is a media claim with limitations corresponding to the limitations of Claim 8 and is rejected under similar rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Gaddam (US 20250252445 A1). Gaddam discloses a LLM using RAG to detect fraudulent transactions.
Erickson (US 20240386041 A1). Erickson discloses a RAG model which uses a user’s profile data to selectively retrieve private data from a vector store.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JON C MEIS whose telephone number is (703)756-1566. The examiner can normally be reached Monday - Thursday, 8:30 am - 5:30 pm EST.
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, Hai Phan can be reached at 571-272-6338. 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.
/JON CHRISTOPHER MEIS/Examiner, Art Unit 2654
/HAI PHAN/Supervisory Patent Examiner, Art Unit 2654
Prosecution Insights