DETAILED ACTION
Status of Claims
Claims 1 and 11 have been amended.
Claims 4 and 14 have been previously cancelled.
Claims 1-3, 5-13, and 15-20 are currently pending and have been considered by the examiner.
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 .
Response to Arguments
101 Rejection:
Applicant’s argument have been considered and have been deemed unpersuasive by the examiner based upon the rationale provided in the following 101 rejection.
Prior Art Rejection:
Applicant asserts that the “typesetting” disclosed by Liang teaches as “word segmentation” and thus is not “truncating…to…count of lines”. However, Applicant merely asserts this as fact without providing any additional rationale. Thus, the examiner must disagree and assert that the text manipulation Liang discloses as “typesetting” or “word segmentation” is functionally equivalent to the BRI of the truncation feature disclosed by the claims. The remainder of applicant’s argument appear to be purely assertive in nature, merely outlining citations made by the previous office action followed by assertion of fact. Thus, the examiner must disagree with them based upon rationale provide both in the previously cited action and the currently provided rejection. Therefore, the examiner must maintain the previously issued prior art rejection.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-3, 5-13, and 15-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
In the instant case, claim 1-3 and 5-10 are directed towards a method and claim 11-13 and 15-20 are directed towards a non-transitory computer readable medium. Therefore, these claims fall within the four statutory categories of invention.
Claim 1 recites the following:
A method comprising:
truncating each software license of a plurality of software licenses from a plurality of license text files that each can contain a software license that can contain a license identifier by loading at most a predefined count of lines of text from the software license into a memory in a computer;
generating, from the plurality of software licenses in the memory, an original plurality of license chunks by splitting each of the plurality of software licenses into one or more license chunks that do not exceed a predefined chunk size;
selecting a selected plurality of license chunks from the original plurality of license chunks that contain at least one of a plurality of strings;
generating a linguistic prompt that contains the selected plurality of license chunks; and
inferentially generating, by a large language model, a license expression from the linguistic prompt.
Wherein the method is performed by one or more computers.
Regarding Step 2A Prong One, the claims recite the abstract idea of performing a mental process. Specifically, the claims recite the limitations underlined above which recite generally claimed method steps which can reasonably be performed by the human mind such as observation and judgement which is grouped within the Mental Processes grouping of abstract ideas in prong one of step 2A of the Alice/Mayo test (See MPEP § 2106.04) because the claims involve the process of performing steps which can reasonably be performed by the human mind. Accordingly, the claims recite an abstract idea (See pages 7, 10, Alice Corporation Pty. Ltd. v. CLS Bank International, et al., US Supreme Court, No. 13-298, June 19, 2014; 2019 Revised Patent Subject Matter Eligibility Guidance, 84 Fed. Reg. 50, 53-54 (January 7, 2019)).
Regarding Step 2A Prong Two, the recited abstract idea is not integrated into a practical application because, when analyzed under prong two of step 2A of the Alice/Mayo test (See MPEP § 2106.04(d)), the additional element(s) of the claim(s) such as a “large language model” and “one or more computers” merely use(s) a computer as a tool to perform an abstract idea. Specifically, the “large language model” and “one or more computers” perform(s) the steps or functions underlined above. The use of a processor/computer as a tool to implement the abstract idea does not integrate the abstract idea into a practical application because it requires no more than a computer performing functions that correspond to acts required to carry out the abstract idea. The additional elements do not involve improvements to the functioning of a computer, or to any other technology or technical field (MPEP 2106.05(a)), the claims do not apply or use the abstract idea to effect a particular treatment or prophylaxis for a disease or medical condition (Vanda Memo), the claims do not apply the abstract idea with, or by use of, a particular machine (MPEP 2106.05(b)), the claims do not effect a transformation or reduction of a particular article to a different state or thing (MPEP 2106.05(c)), and the claims do not apply or use the abstract idea in some other meaningful way beyond generally linking the use of the abstract idea to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolize the exception (MPEP 2106.05(e) and Vanda Memo). Therefore, the claims do not, for example, purport to improve the functioning of a computer. Nor do they effect an improvement in any other technology or technical field. Accordingly, the additional elements do not impose any meaningful limits on practicing the abstract idea, and the claims are directed to an abstract idea.
The claim(s) do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, when analyzed under step 2B of the Alice/Mayo test (See MPEP § 2106.05), the additional element(s) of a “large language model” and “one or more computers” amounts to no more than using a computer or processor to automate and/or implement the abstract idea. As discussed above, taking the claim elements separately, the “large language model” and “one or more computers” perform(s) the steps or functions underlined above. These functions correspond to the actions required to perform the abstract idea. Viewed as a whole, the combination of elements recited in the claims merely recite performing a mental process. Therefore, the use of these additional elements does no more than employ the computer as a tool to automate and/or implement the abstract idea. The use of a computer or processor to merely automate and/or implement the abstract idea cannot provide significantly more than the abstract idea itself (MPEP 2106.05(I)(A)(f) & (h)). Therefore, the claim is not patent eligible.
Dependent claims 2-3, 5-10, 12-13, and 15-20 further describe the recited abstract idea. The dependent claims do not include additional elements that integrate the abstract idea into a practical application or that provide significantly more than the abstract idea. Specifically:
Claims 2, 5-8, 12, and 15-18 merely further describe the data being processed when performing the recited abstract idea.
Claims 3 and 13 merely further describes the steps directed towards the recited abstract idea.
Claims 9-10 and 19-20 recite additional limitations which are also directed towards the recited abstract idea.
Therefore, as the dependent claims do not include additional elements that integrate the abstract idea into a practical application nor provide significantly more than the abstract idea, the dependent claims are also not patent eligible.
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, 5-9, 11-13, and 15-19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Longoni (US 20250245218 A1) in view of Raffoul et al. (US 20210090190 A1) in further view of Liang (CN 117195886 A1).
Regarding Claims 1 and 11, Longoni discloses:
A method comprising:
Truncating each text file of a plurality of text files from a plurality of files by loading at most a predefined amount of text from the text files into a memory in a computer (See Longoni: Para. [0049] – “The process 200 includes a data ingestion phase 205. At for instance, a compute device (e.g., AI ingest engine or compute device 101 of FIG. 1) of the RAG AI system can receive multiple data. The data source can be a document (PDF, Word, text), an Excel file, or simply a text file containing ASCII text. In some cases, an appropriate document loader needs to be provisioned to be able to open the specific file type and extract: metadata such as filename, page number, document title if available, and most importantly the text content to be analyzed.”);
Generating, from the plurality of text files in memory, chunks by splitting each of the data elements in to chunks that do not exceed a predefined chunk size (See Longoni: para. [0034] – “Once the data artifacts 110 are converted and/or encoded to the standard data type (e.g., PDF), a tokenization process can be performed by the processor 102. In some implementations, tokenization can be required to subdivide natural language identifiers (e.g., text) into “sentence particles” that can be then turned into vectors 113 (or vector representations). The tokenization process can be a function of “chunk size” and “chunk overlap.” For instance, the chunk size can represent a fixed size of paragraphs being extracted from the data artifacts 110 (or encoded documents). The chunk overlap can represent an overlap between paragraphs, for example when a paragraph continues to a next page. In some implementations, a predetermined parameter for chunk size and chunk overlap in the tokenization process to generate the tokens 111. For instance, the tokens 111 can include, for example, 2000 tokens for the chunk size and 200 tokens for overlap. In some cases, the parameters for the chunk size and chunk overlap of the tokens 111 can drive the context size of semantic search and can be used to improve the accuracy of the embedding model 114 in answering queries from the user compute device 121 (or multiple queries from multiple user compute devices).”);
selecting a selected plurality of chunks from the original plurality of chunks that contain at least one of a plurality of strings (See Longoni: Para. [0048] – “A query engine using the HTTP POST method can be designed to communicate with the API inferencing server. Standard techniques can be employed to achieve this task. A module can then be designed that utilizes the metadata obtained from the vector store database to extract the document page and create an image that can be rendered on the screen. This approach allows the user to quickly identify the data sources that the AI used in generating the answer. A chat memory feature can be designed by modifying a prompt passed to the AI (e.g., LLM 116) for processing and to be displayed on the user compute device 121. The chat memory can enhance the user's experience by providing a conversational behavior, where the AI can respond to follow up questions in a quasi-human fashion. The prompt to direct the AI can be modified to answer the question by using the user's query, the context found via semantic search, as well as the chat history stored in an array of strings.”);
generating a linguistic prompt that contains the selected plurality of chunks (See Longoni: Para. [0035] – “The vectors 113 (or embeddings) can include transformed text of the data artifacts 110 via the embedding model 114. In some cases, the vectors 113 can include numerical representations of objects, such as words, phrases, or data artifacts 110, in a vector space. In some cases, the vectors 113 can be 1-D vectors. In some implementations, the embedding model 114 can include a vocabulary that maps a word such as “family” to an integer number. Each word in a sentence can then transformed into an integer, creating a string of integer numbers. This string can then be “embedded” into the 1-D vector.”); and
inferentially generating, by a large language model, an expression from the linguistic prompt (See Longoni: Para. [0041] – “In the inferencing phase, the embedding model 114 (that is trained in the training phase) can receive at least one document (a document not among the set of documents (e.g., data artifacts 110) used in the training phase) and annotates and/or classifies words and/or paragraphs of the document. Because the inferencing phase is performed using the set model parameters that were already optimized during the training phase, the inferencing phase is computationally efficient.”).
However, Longoni fails to explicitly disclose:
Wherein the text files comprise a plurality of software licenses which each contain a license identifier
However, in a similar field of endeavor, Raffoul discloses:
Wherein the text files comprise a plurality of software licenses which each contain a license identifier (See Raffoul: Para. [0034] – “For transaction data files, such as contracts and other agreements, users can enter electronic copies of these data files into the system. These can be added to the system's database 60. As examples, a user can enter all transaction data files (e.g., employment, software license, sales or channel agreements) into the system. The system can then apply optical character recognition to convert the images of the agreements into computer-readable text and then the system can apply machine learning and/or artificial intelligence-based methods and subsystems to analyze the agreements. These methods and subsystems may include techniques related to natural language processing (NLP), using artificial intelligence models trained on legal documents by legal experts. Additionally, of course, optical character recognition may not be required as a separate step: for instance, contracts may be entered as text files that do not require special conversion. Alternatively, some artificial intelligence models may be able to combine character-recognition tasks and semantic parsing tasks, rather than requiring separate processes.”; See Raffoul: Para. [0037] – “The documentation analysis module 50 then identifies a presence or an absence of at least one transaction term in the analyzed data file.”)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to substitute the generic text files disclosed by Longini for the text files comprising software licenses as disclose by Raffoul yielding the predictable result of an increase in the economic viability of the invention by expanding the types of data which can be processed by the invention.
However, the combination of Longini and Raffoul fails to explicitly disclose:
including truncating each of the plurality of software licenses to a predefined count of lines of text.
However, in a similar field of endeavor, Liang discloses:
including truncating text documents to a predefined count of lines of text (See Liang: pgs 9-10 – “In this embodiment, the text truncation processing can be performed at the position where the targetline-feed character exists in the fourth specified text to complete the line … In this embodiment, the above adjustment rule is a pre-constructed rule for checking and adjusting the generated text after the line-feed processing, so that the text after the line-feed processing can meet the preset text typesetting specification. The content of the text typesetting specification is not specifically limited, and can be determined according to the actual text processing requirement. adjusting the third text based on the adjusting rule so as to obtain a fourth text in accordance with thepreset expected condition.”).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify the transaction method disclosed by Longini and Raffoul for the truncation method based upon a predefined target line-feed meeting a preset text typesetting specification yielding the predictable result of an increase in the processing speed of the invention by ensuring that the licenses are not truncated improperly.
Regarding Claims 2 and 12, the combination discloses:
wherein the linguistic prompt contains at least one selected from a group consisting of:the license identifiers in the plurality of software licenses, a binary logical operator that the license expression contains, a binary logical operator that the license expression does not contain, and a natural language explanation of available logical operators that could be included in the license expression (See Longoni: Para. [0035] – “The vectors 113 (or embeddings) can include transformed text of the data artifacts 110 via the embedding model 114. In some cases, the vectors 113 can include numerical representations of objects, such as words, phrases, or data artifacts 110, in a vector space. In some cases, the vectors 113 can be 1-D vectors. In some implementations, the embedding model 114 can include a vocabulary that maps a word such as “family” to an integer number. Each word in a sentence can then transformed into an integer, creating a string of integer numbers. This string can then be “embedded” into the 1-D vector.”).
Regarding Claims 3 and 13, the combination discloses:
wherein said inferentially generating comprises inferentially generating a natural language explanation of the license expression (See Longoni: Abstract – “The hash functions and encoded documents are stored in a first database. The processor is caused to tokenize the encoded artifacts, to produce tokens associated with natural-language identifiers extracted from the encoded artifacts”).
Regarding Claims 4 and 14, the combination discloses:
further comprising truncating each of the plurality of software licenses to a predefined count of lines of text (See Longoni: para. [0034] – “The tokenization process can be a function of “chunk size” and “chunk overlap.” For instance, the chunk size can represent a fixed size of paragraphs being extracted from the data artifacts 110 (or encoded documents). The chunk overlap can represent an overlap between paragraphs, for example when a paragraph continues to a next page”).
Regarding Claims 5 and 15, the combination discloses:
wherein the predefined count of lines of text does not exceed twenty (See Longoni: Para. [0034] – “The tokenization process can be a function of “chunk size” and “chunk overlap.” For instance, the chunk size can represent a fixed size of paragraphs being extracted from the data artifacts 110 (or encoded documents). The chunk overlap can represent an overlap between paragraphs, for example when a paragraph continues to a next page. In some implementations, a predetermined parameter for chunk size and chunk overlap in the tokenization process to generate the tokens 111” – It is clear to one of ordinary skill in the art that the predetermined parameter can be any value including twenty lines of text/the equivalent in chunks).
Regarding Claims 6 and 16, the combination discloses:
wherein the license expression does not contain a particular license identifier in the plurality of software licenses (See Longoni: Para. [0041] – “In the inferencing phase, the embedding model 114 (that is trained in the training phase) can receive at least one document (a document not among the set of documents (e.g., data artifacts 110) used in the training phase) and annotates and/or classifies words and/or paragraphs of the document. Because the inferencing phase is performed using the set model parameters that were already optimized during the training phase, the inferencing phase is computationally efficient.”).
Regarding Claims 7 and 17, the combination discloses:
wherein the plurality of strings includes the license identifiers in the plurality of software licenses (See Longoni: Para. [0041] – “In the inferencing phase, the embedding model 114 (that is trained in the training phase) can receive at least one document (a document not among the set of documents (e.g., data artifacts 110) used in the training phase) and annotates and/or classifies words and/or paragraphs of the document. Because the inferencing phase is performed using the set model parameters that were already optimized during the training phase, the inferencing phase is computationally efficient.”; See Raffoul: Para. [0037] – “The documentation analysis module 50 then identifies a presence or an absence of at least one transaction term in the analyzed data file.”).
Regarding Claims 8 and 18, the combination discloses:
Wherein each license text file of the plurality of license text files has a filename; multiple license text files of the plurality of license text files have a same filename (See Longoni: Para. [0033] – “The first database 118 can be used to store ingested data (e.g., data artifacts 110), encoded documents (e.g., converted PDF documents), and the hash functions 112. For instance, the first database 118 can store file names, hash values, dates and/or timestamps when the files were added, file size, and/or the like.”);
the method further comprises in a filesystem or software repository, selecting the plurality of license text files by selecting files that have a filename part in a predefined plurality of filename parts (See Longoni: Para. [0060] – “As shown in FIG. 7, the user can be presented via the AI query interface data of ingested data including number of files in a collection (e.g., collection of data based on specific topics/categories), file size, collection size, file name, and/or the like. As shown in FIG. 8, a user-friendly chatbot like interaction can be seen that provides answers to user queries.”).
Regarding Claims 9 and 19, the combination discloses:
further comprising prepending the license identifier onto each of the plurality of software licenses (See Longoni: Para. [0053] – “At a data ingestion side, and at 303, a compute device (e.g., AI ingest engine or compute device 101 of FIG. 1) can receive a document 301 to be loaded and parsed for storage in a database 304 (e.g., first database 118 of FIG. 1). At 305, the parsed document can be embedded into vectors 306, which can be finetuned. The vector 306 can be stored in a vector database (e.g., second database 119 of FIG. 1).”).
Claim(s) 10 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Longoni in view of Roffoul in further view of Liang and McBrearty et al. (US 20080109712 A1)
Regarding Claims 10 and 20, the combination of Longoni, Roffoul, and Liang discloses the method of claim 1 but fails to explicitly disclose:
further comprising in the plurality of software licenses, replacing a plurality of uniform resource locators (URLs) with a predefined replacement string.
However, in a similar field of endeavor, McBrearty discloses:
further comprising in the plurality of software licenses, replacing a plurality of uniform resource locators (URLs) with a predefined replacement string (See McBrearty: Para. [0014] – “The present invention supports the substitution of a textual string derived from a web page title in place of a uniform resource locator (URL) in the text of a document”).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to implement the method of replacing URL’s within strings disclosed by McBrearty to replace any URL’s contained with the text files disclosed by Longoni, Roffoul, and Liang yielding the predictable result of a more efficient system by preventing the large language model from needing to be trained to handle the random strings of characters which construct URL’s.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NICHOLAS K PHAN whose telephone number is (571)272-6748. The examiner can normally be reached M-F 1 pm-9 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, Neha Patel can be reached at 571-270-1492. 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.
/NICHOLAS K PHAN/Examiner, Art Unit 3699
Prosecution Insights