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 .
Response to Arguments/Amendments
Applicant's cancellation of claims 14 and 28 overcome the 35 U.S.C. 112(b) rejection and the rejection has been withdrawn.
Applicant's arguments with respect to 35 U.S.C. 101 in regard to claims 1-28 have been considered, however are not found to be persuasive due to the following reasons. Examiner respectfully disagrees with Applicant arguments because under Step 1 under Alice/Mayo analysis, the claims remain directed to evaluating a debate argument, judging its persuasiveness and impact, and assigning classifications and scores, activities that are abstract mental processes. Although continuous tokens and backpropagation cannot practically be performed is a mathematical operation. These limitations merely use ML to automate the abstract evaluation.
The additional limitations also do not integrate the abstract idea into a practical application or provide an inventive concept under Alice/Mayo Step 2. The claims recite fine-tuning an LLM using tokens, backpropagation, and class labels, but they do not claim a particular model architecture or a specific improvement to computer speed, memory, efficiency, or operation. The output remains informational classifications and scores. Therefore, the Applicant’s unsupported assertion that the method provides an unconventional technological improvement does not overcome the 101 rejection.
Applicant's arguments with respect to 35 U.S.C. 103 rejection of claims 1 and 15 have been considered and found persuasive, and the rejection has been withdrawn. See detailed reason for allowance below.
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 1-13 and 15-27 are rejected under 35 U.S.C. 101.
Claims 1 and 15 are directed to an Abstract Idea because, under Step 1, prong one, the claims recite the abstract ideas of evaluating information, forming judgements, and assigning labels or numerical scores. Defining persona characteristics and judging the persuasiveness and impact of a debate argument are activities that people can perform mentally. Updating continuous tokens through backpropagation also involves mathematical calculations.
Under prong two, the claims do not integrate these abstract ideas into a practical application. The LLM, prompts, pretrained models, and tokens are used primarily to automate the collection, organization, and evaluation of information. The claimed result is still informational: classifications and quantitative scores concerning the persuasiveness and impact of an argument.
Under Step 2B, the additional limitations do not provide an inventive concept. They apply conventional computer and ML operations without claiming a specific improvement to computer operation or producing a concrete technological result.
The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the claims are (i) mere instructions to implement the idea on a computer, and/or (ii) recitation of generic computer structure that serves to perform generic computer functions that are well-understood, routine, and conventional activities previously known to the pertinent industry. Viewed as a whole, these additional claim element(s) do not provide meaningful limitation(s) to transform the abstract idea into a patent eligible application of the abstract idea such that the claim(s) amounts to significantly more than the abstract idea itself. Therefore, the claim(s) are rejected under 35 U.S.C. 101 as being directed to non-statutory subject matter. There is further no improvement to the computing device.
Dependent claims 2-13 and 16-27 further recite an abstract idea performable by a human and do not amount to significantly more than the abstract idea as they do not provide steps other than what is conventionally known.
Claims 2 and 16, adding a persona stance only adds another type of opinion to the same abstract idea of collecting and evaluating information about arguments, using generic AI tools rather than a specific technological improvement.
Claims 3 and 17, adding a persona argument that supports the stance only adds more argumentative content to the same abstract idea, and does not improve how the computer or model itself works.
Claims 4 and 18, adding a persona character merely adds more personal-trait information to the same abstract idea of profiling and evaluating arguments, implemented with generic computer and AI functions.
Claims 5 and 19, adding a persona intent only adds another category of human-related information to the same abstract idea, without reciting a specific improvement in computer technology.
Claims 6-11 and 20-25, scoring is still just an evaluation or judgment about information, which remains part of the same abstract idea.
Claims 12 and 26, adding psychological principles, prompt tuning, and linguistic or discourse features only adds more rules and considerations for analyzing and presenting arguments, carried out with generic Al tools, rather than a specific improvement in computer functionality.
Claims 13 and 27, adding factors such as speaker tone, body language, and perceived credibility only adds more human-persuasion considerations to the same abstract idea, without changing how the computer operates.
Allowable Subject Matter
Claims 1-13 and 15-27 would be allowable if the Applicant can overcome the 101 Abstract Idea set forth.
The following is a statement of reasons for the indication of allowable subject matter:
Al-Khatib et al. (“Exploiting Personal Characteristics of Debaters for Predicting Persuasiveness”; 2020) teaches automatically evaluating online-debate arguments for persuasiveness using persona-related characteristics. It models “debaters prior beliefs, interests, and personality traits based on their previous activity” and reports that these characteristics improve prediction of “argument persuasiveness” and “resistance to persuasion” ([Abstract]). The reference defines the classification task as predicting , “given a debate topic and an argument regarding it,” whether the argument “is able to change the stance of an opponent” ([3.]). It trains on samples containing the argument and author characteristics with a “binary target of whether a delta was awarded” and reports quantitative accuracy results for interests, beliefs, and personality traits ([5.1-5.2] [Table 3]).
Xu et al. (“ExpertPrompting: Instructing Large Language Models to be Distinguished Experts”; May 2023) teaches using a tailored expert persona to improve an LLM’s response to a particular instruction. Xu uses “in-context learning to automatically synthesize detailed and customized descriptions of the expert identity for each specific instruction” and asks the LLM to answer “conditioned on such agent background” ([Abstract]). The expert description is “customized to each specific instruction” and “detailed and comprehensive,” while instruction-expert examples are places in the prompt to generate a suitable expert identity ([2.1] [eq. 2] [Fig. 6]). Xu then “pairs each expert identity” with the original instruction to obtain an augmented answer ([2.2] [eq. 3] [Fig. 7]). These teachings are relevant to the claimed persona-knowledge prompt, in-context demonstration, eliciting persona knowledge from an LLM and combining that knowledge with the task input.
Lester et al. (“The Power of Scale for Parameter-Efficient Prompt Tuning”; 2021) teaches parameter-efficient prompt tuning for adapting a LM to classification tasks. Lester explains that “soft prompts are learned through back-propagation” ([Abstract]). For classification, Lester “casts all tasks as text generation” and provides that “Y is a sequence of tokens that represent a class label” ([2.]). Lester further represents the soft prompt as a continuous parameter matrix and trains it though backpropagation while updating only the prompt parameters ([2.]). These teachings are relevant to the claimed continuous prompt tokens, backpropagation updates, conditional-generation classification, and output token corresponding to class labels.
The difference between the prior art and the claimed invention is that Al-Khatib, Xu nor Lester explicitly teach utilizing a knowledge-aligned prompt to formulate the classification problems as conditional generation tasks to generate learnable continuous tokens; updating the learnable continuous tokens by backpropagation to generate output tokens; and mapping each of the generated output tokens to a class label in fine-tuning the debating LLM.
Therefore, it would not have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to modify the teachings of Al-Khatib, Xu and Lester to include utilizing a knowledge-aligned prompt to formulate the classification problems as conditional generation tasks to generate learnable continuous tokens; updating the learnable continuous tokens by backpropagation to generate output tokens; and mapping each of the generated output tokens to a class label in fine-tuning the debating LLM.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHREYANS A PATEL whose telephone number is (571)270-0689. The examiner can normally be reached Monday-Friday 8am-5pm PST.
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, Pierre Desir can be reached at 571-272-7799. 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.
SHREYANS A. PATEL
Primary Examiner
Art Unit 2653
/SHREYANS A PATEL/Examiner, Art Unit 2659
Prosecution Insights