CTNF 18/491,877 CTNF 102121 Notice of Pre-AIA or AIA Status 07-03-aia AIA 15-10-aia The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA. Information Disclosure Statement The information disclosure statement (IDS) submitted on October 30, 2024 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner. Claim Rejections - 35 USC § 101 07-04-01 AIA 07-04 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 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. Below is a claim-by-claim analysis. Claim 1, 11, 20: Step 1: Recites a method (claim 1), a system (claim 11) and a manufacture (claim 20). Therefore, they are directed to the statutory categories of invention. Step 2A Prong 1: The claims recite: generate a first plurality of inferred labels for the first plurality of augmented training examples Generating inferred labels (i.e. classifying) is a mental process that can be done in one’s head. predict the first plurality of inferred labels for the first plurality of augmented training examples Predicting the inferred labels (the classifications) is a mental process that can be done in one’s head. generate a second plurality of inferred labels for the second plurality of unlabeled training examples Generating inferred labels (i.e. classifying) is a mental process that can be done in one’s head. predict the second plurality of inferred labels for the second plurality of unlabeled training examples Predicting the inferred labels (the classifications) is a mental process that can be done in one’s head. Step 2A Prong 2 : The judicial exception is not integrated into a practical application. The remaining limitations of the claims are either directed at mere data gathering (“obtaining a first plurality of unlabeled training examples…”), insignificant extra-solution activity (“augmenting the first plurality of queries…”), or are further components of the abstract idea (“training a second processing model to predict …”, training a third processing model to predict …”). Step 2B: The claim does not contain significantly more than the judicial exception. The analysis mirrors the analysis of step 2A prong 2. The claim as a whole is directed to the abstract of classifying queries through multiple iterations, and does not contain more that amounts to an inventive concept. Claim 2: Step 1: Recites a method. Therefore, it is directed to the statutory categories of invention. Step 2A Prong 1: The claim recites the abstract idea it inherits from the claim it depends on. Step 2A Prong 2: The judicial exception is not integrated into a practical application. The remaining limitation of the claim is directed at additional components of the mere data gathering (“…compromises queries of technical terminologies”). Step 2B: The claim does not contain significantly more than the judicial exception. The analysis mirrors the analysis of step 2A prong 2. Claim 3, 12: Step 1: Recites a method (claim 3) and a system (claim 12). Therefore, they are directed to the statutory categories of invention. Step 2A Prong 1: The claim recites the abstract idea it inherits from the claim it depends on. Step 2A Prong 2: The judicial exception is not integrated into a practical application. The remaining limitation of the claim is directed at additional components of the insignificant extra-solution activity (“…compromises URLs of documents retrieved”). Step 2B: The claim does not contain significantly more than the judicial exception. The analysis mirrors the analysis of step 2A prong 2. Claim 4, 13: Step 1: Recites a method (claim 4) and a system (claim 13). Therefore, they are directed to the statutory categories of invention. Step 2A Prong 1: The claim recites the abstract idea it inherits from the claim it depends on. Step 2A Prong 2: The judicial exception is not integrated into a practical application. The remaining limitation of the claim is directed at additional components of the insignificant extra-solution activity (“…compromises titles of document retrieved”). Step 2B: The claim does not contain significantly more than the judicial exception. The analysis mirrors the analysis of step 2A prong 2. Claim 5, 14: Step 1: Recites a method (claim 5) and a system (claim 14). Therefore, they are directed to the statutory categories of invention. Step 2A Prong 1: The claim recites the abstract idea it inherits from the claim it depends on. Step 2A Prong 2: The judicial exception is not integrated into a practical application. The remaining limitation of the claim is directed at additional components of the application of the judicial exception (“…based on the first plurality of queries exclusive of the one or more additional feature annotations”). Step 2B: The claim does not contain significantly more than the judicial exception. The analysis mirrors the analysis of step 2A prong 2. Claim 6, 15: Step 1: Recites a method (claim 6) and a system (claim 15). Therefore, they are directed to the statutory categories of invention. Step 2A Prong 1: The claim recites the abstract idea it inherits from the claim it depends on. Step 2A Prong 2: The judicial exception is not integrated into a practical application. The remaining limitation of the claim is directed at additional components of the application of the judicial exception (“second query processing model comprises a same number of parameters as the first query processing model”). Step 2B: The claim does not contain significantly more than the judicial exception. The analysis mirrors the analysis of step 2A prong 2. Claim 7, 16: Step 1: Recites a method (claim 7) and a system (claim 16). Therefore, they are directed to the statutory categories of invention. Step 2A Prong 1: The claim recites the abstract idea it inherits from the claim it depends on. Step 2A Prong 2: The judicial exception is not integrated into a practical application. The remaining limitation of the claim is directed at additional components of the mere data gathering (“second plurality of unlabeled training examples comprises a larger number of training examples than first…”). Step 2B: The claim does not contain significantly more than the judicial exception. The analysis mirrors the analysis of step 2A prong 2. Claim 8, 17: Step 1: Recites a method (claim 8) and a system (claim 17). Therefore, they are directed to the statutory categories of invention. Step 2A Prong 1: The claim recites the abstract idea it inherits from the claim it depends on. Step 2A Prong 2: The judicial exception is not integrated into a practical application. The remaining limitation of the claim is directed at additional components of the application of the judicial exception (“third query processing model comprises a smaller number of parameters than the second…”). Step 2B: The claim does not contain significantly more than the judicial exception. The analysis mirrors the analysis of step 2A prong 2. Claim 9, 18: Step 1: Recites a method (claim 9) and a system (claim 18). Therefore, they are directed to the statutory categories of invention. Step 2A Prong 1: The claim recites the abstract idea it inherits from the claim it depends on. Step 2A Prong 2: The judicial exception is not integrated into a practical application. The remaining limitation of the claim is directed at additional components of the mere data gathering (“further comprising obtaining a first plurality of labeled training examples…”). Step 2B: The claim does not contain significantly more than the judicial exception. The analysis mirrors the analysis of step 2A prong 2. Claim 10, 19: Step 1: Recites a method (claim 10) and a system (claim 19). Therefore, they are directed to the statutory categories of invention. Step 2A Prong 1: The claim recites the abstract idea it inherits from the claim it depends on. Step 2A Prong 2: The judicial exception is not integrated into a practical application. The remaining limitation of the claim is directed at additional components of the mere data gathering (“first plurality of labeled training examples are labeled with query intent class”). Step 2B: The claim does not contain significantly more than the judicial exception. The analysis mirrors the analysis of step 2A prong 2. Claim Rejections - 35 USC § 103 07-20-aia AIA 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. 07-21-aia AIA Claim s 1, 3, 6, 8-9, 11-12, 15, 17-18. And 20 are rejected under 35 U.S.C. 103 as being unpatentable over Yang et al. (“Model Compression with Two-stage Multi-teacher Knowledge Distillation for Web Question Answering System”, 2020), in view of Mirzadeh et al. (“Improved Knowledge Distillation via Teacher Assistant”, 2020, listed on IDS filed 10/30/2024 ), and in further view Abdi et al. (“US 11132583“, 2021) . Regarding claim 1, Yang discloses a method comprising: obtaining a first plurality of unlabeled training examples the first plurality of unlabeled training examples respectively comprising a first plurality of queries ( Page 691, Column 2, Paragraph 5, “in the first stage… leverage large-scale unlabeled question-passage pairs ); augmenting the first plurality of queries with one or more additional feature annotations to generate a first plurality of augmented training examples ( Page 693, Column 1, Paragraph 3, “each question, top 10 relevant documents are returned by the commercial search engine to form <Question, URL> pairs” ); processing the first plurality of augmented training examples with a first query processing model to respectively generate a first plurality of inferred labels for the first plurality of augmented training examples ( Page 694, Column 1, Paragraph 3, “At the first stage… learn the generalized natural language inference capability from the unlabeled data with soft labels” ); training a second query processing model to predict the first plurality of inferred labels for the first plurality of augmented training examples ( Page 691, Column 1, Paragraph 2, “when training the student model… also feed the output score from the teacher model as a secondary soft label” ); obtaining a second plurality of unlabeled training examples, the second plurality of unlabeled training examples respectively comprising a second plurality of queries… ( Page 694, Column 2, Paragraph 2, “sampled 4 million (named base dataset) and 40 million (named large dataset) as the pre-training data” ); processing the second plurality of unlabeled training examples ( Page 695, Column 2, Paragraph 3, 4.4 Parameter Settings, “while TMKD large is pre-trained using CommQA-Unlabeled large corpus” ) 1 with the second query processing model to respectively generate a second plurality of inferred labels for the second plurality of unlabeled training examples ( Page 693 , Column 2, Paragraph 1, “jointly learns from multiple teachers simultaneously during training, as well as generates final prediction output” ); Yang fails to teach each of the second plurality of unlabeled training examples comprising a smaller number of features than the first plurality of augmented training examples and training a third query processing model to predict the second plurality of inferred labels for the second plurality of unlabeled training examples. However, Mirzadeh discloses training a third… model to predict the second plurality of inferred labels… 2 ( Page 5191, Column 1, Paragraph 1, “we show that the student network performance degrades when the gap between student and teacher is large…To acknowledge…employs an intermediate-sized network (teacher assistant) to bridge the gap” ), and Abdi discloses each of the second plurality of… 3 training examples comprising a smaller number of features than the first plurality of augmented training examples ( Claim 16 , “grouping data of a first training set into a plurality of clusters, each of the plurality of clusters including multiple types of data having a common characteristic with one another… the second training set having a fewer number of types of data than a number of types of data in the cluster” ). Yang, Mirzadeh, and Abdi are considered analogous to the invention because all are in the field of systems for the transfer of learning in neural networks. Therefore, it would’ve been obvious to one of ordinary skill of the art before the effective filing date of the invention to have modified Yang to incorporate the teachings of Mirzadeh and Abdi. Combining with Mirzadeh helps to bridge the knowledge gap between the student and teacher models ( see abstract of Mirzadeh ), and combining with Abdi helps to quicken processing ( see Paragraph 2 of Abdi ). Regarding claim 3, Yang discloses a method wherein the one or more additional feature annotations comprises URLs of documents retrieved for the first plurality of queries ( Page 694, Column 1, Paragraph 1, “each question, top 10 relevant documents are returned by the commercial search engine to form <Question, URL> pairs” ). Regarding claim 6, Mirzadeh discloses a method wherein the second… model comprises a same number of parameters as the first… 4 model ( Page 5192, Column 2, Paragraph 2, “argue that distillation can even work when the teacher and student are made by the same network architecture” ). Mirzadeh is considered analogous to the invention because it concerns knowledge distillation in neural network systems. Therefore, it would’ve been obvious to one of ordinary skill of the art before the effective filing date of the invention to have modified Yang to incorporate the teachings of Mirzadeh Doing so helps to bridge the knowledge gap between the student and teacher models and makes learning transfer more effective ( see abstract of Mirzadeh ). Regarding claim 8, Mirzadeh discloses a method wherein the third… processing model comprises a smaller number of parameters than the second… 5 processing model ( Page 5193, Column 2, Paragraph 5, “The teacher assistant (TA) lies somewhere in between teacher and student in terms of size 6 or capacity” ). Mirzadeh is considered analogous to the invention because it concerns knowledge distillation in neural network systems. Therefore, it would’ve been obvious to one of ordinary skill of the art before the effective filing date of the invention to have modified Yang to incorporate the teachings of Mirzadeh Doing so helps to bridge the knowledge gap between the student and teacher models and makes learning transfer more effective ( see abstract of Mirzadeh ). Regarding claim 9, Yang discloses a method further comprising obtaining a first plurality of labeled training examples, the first plurality of labeled training examples respectively comprising a first plurality of queries ( Page 693, Column 1, Paragraph 8, “we use both the golden label (i.e. ground-truth knowledge of T 0 ) on the task specific corpus and the soft labels of T 1 -T N (i.e. pseudo ground truth knowledge) on the same corpus to jointly fine-tune to get an enhanced student model” ). Claim 11 is a system claim corresponding to method claim 1 and is rejected for the same reasons as given in the rejection of that claim. Claim 12 is a system claim corresponding to method claim 3 and is rejected for the same reasons as given in the rejection of that claim. Claim 15 is a system claim corresponding to method claim 6 and is rejected for the same reasons as given in the rejection of that claim. Claim 17 is a system claim corresponding to method claim 8 and is rejected for the same reasons as given in the rejection of that claim. Claim 18 is a system claim corresponding to method claim 9 and is rejected for the same reasons as given in the rejection of that claim. Claim 20 is a non-transitory computer-readable medium claim corresponding to method claim 1 and is rejected for the same reasons as given in the rejection of that claim . 07-21-aia AIA Claim 2 is rejected under 35 U.S.C. 103 as being unpatentable over Yang et al. (“Model Compression with Two-stage Multi-teacher Knowledge Distillation for Web Question Answering System”, 2020), in view of Mirzadeh et al. (“Improved Knowledge Distillation via Teacher Assistant”, 2020), and in further view Abdi et al. (“US 11132583“, 2021) and Rinaldi et al. (“The Role of Technical Terminology in Question Answering”, 2003) . Yang, Mirzadeh, Abdi fail to disclose the further limitations of the claim. However, Rinaldi discloses a method wherein the first plurality of queries comprises queries of technical terminologies ( Page 2, Paragraph 3-4 , “role terminology plays in technical domains has long been recognized…One area where such requirements are particularly pressing is that of ‘Question Answering’ (QA) … approach taken by QA systems is to allow a user to ask a query” ). Rinaldi is considered analogous to the invention because both concern systems that process queries. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified Yang, Mirzadeh, and Abdi to incorporate the teachings of Rinaldi. Doing so allows a system capable of deeper linguistic analysis than otherwise would be capable of ( see Page 1, Paragraph 1 of Introduction of Rinaldi ) . 07-21-aia AIA Claim 4, 13 are rejected under 35 U.S.C. 103 as being unpatentable over Yang et al. (“Model Compression with Two-stage Multi-teacher Knowledge Distillation for Web Question Answering System”, 2020), in view of Mirzadeh et al. (“Improved Knowledge Distillation via Teacher Assistant”, 2020), and in further view Abdi et al. (“US 11132583“, 2021) and Komeili et al. (“Internet-Augmented Dialogue Generation”, 2022) . Regarding claim 4, Yang, Mirzadeh, Abdi fail to disclose the further limitations of the claim. However, Komeili discloses a method wherein the one or more additional feature annotations comprises titles of documents retrieved for the first plurality of queries ( Page 8463, Column 1, Paragraph 4, Search Engine-Augmented Generation, “In addition, we can also consider if the URL is from English Wikipedia, in that case we can extract the page title from the URL and look up its corresponding page” ). Komeili is considered analogous to the invention because both concern retrieval augmented systems that involve working with user queries. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified Yang, Mirzadeh, and Abdi to incorporate the teachings of Komeili. Doing so allows for a system that is grounded and up-to-date, and reduces the likelihood of hallucination ( see Abstract of Komeili ). Claim 13 is a system claim corresponding to method claim 4 and is rejected for the same reasons as given in the rejection of that claim . 07-21-aia AIA Claim 5, 14 are rejected under 35 U.S.C. 103 as being unpatentable over Yang et al. (“Model Compression with Two-stage Multi-teacher Knowledge Distillation for Web Question Answering System”, 2020), in view of Mirzadeh et al. (“Improved Knowledge Distillation via Teacher Assistant”, 2020), and in further view Abdi et al. (“US 11132583“, 2021) and Sivakumar et al. (“US 20210287084”) . Regarding claim 5, Yang discloses training the second query processing model to predict the first plurality of inferred labels for the first plurality of augmented training examples ( see claim 1 analysis ). Yang, Mirzadeh, Abdi fail to disclose the further limitations of the claim. However, Sivakumar discloses training the… model to predict the first plurality of inferred labels based on the first plurality… 7 exclusive of the one or more additional feature annotations ( Paragraph 71, “another instance of the model may be trained using a non-augmented sample set” ). Sivakumar is considered analogous to the invention because both concern optimizations to machine learning models. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified Yang, Mirzadeh, and Abdi to incorporate the teachings of Sivakumar. Doing so allows for a more robust system as training with both augmented and non-augmented datasets lets you adjust how the augmentation is weighed ( see Paragraph 74 of Sivakumar ). Claim 14 is a system claim corresponding to method claim 5 and is rejected for the same reasons as given in the rejection of that claim . 07-21-aia AIA Claim 7, 16 are rejected under 35 U.S.C. 103 as being unpatentable over Yang et al. (“Model Compression with Two-stage Multi-teacher Knowledge Distillation for Web Question Answering System”, 2020), in view of Mirzadeh et al. (“Improved Knowledge Distillation via Teacher Assistant”, 2020), and in further view Abdi et al. (“US 11132583“, 2021) and Mou et al. (“WO 2021159769”) . Regarding claim 7, Yang, Mirzadeh, Abdi fail to disclose the further limitations of the claim. However, Mou disclose a method wherein the second plurality of unlabeled training examples comprises a larger number of training examples than the first plurality of unlabeled training examples ( Paragraph 301, “training data includes both labeled sample data and unlabeled sample data… the number of samples included in the first training data set is less than the number of samples included in the second training data set” ). Mou is considered analogous to the invention because both concern the field of artificial intelligence. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified Yang, Mirzadeh, and Abdi to incorporate the teachings of Mou. Doing so ensures that the deep features of the data are learned ( see Paragraph 301 of Mou ). Claim 16 is a system claim corresponding to method claim 7 and is rejected for the same reasons as given in the rejection of that claim . 07-21-aia AIA Claim 10, 19 is rejected under 35 U.S.C. 103 as being unpatentable over Mirzadeh et al. (“Improved Knowledge Distillation via Teacher Assistant”, 2020), in view of Mirzadeh et al. (“Improved Knowledge Distillation via Teacher Assistant”, 2020), and in further view Abdi et al. (“US 11132583“, 2021) and Shah et al. (”US 20220129633”) . Regarding claim 10, Yang, Mirzadeh, and Abdi fail to disclose the further limitations of the claim. However, Shah discloses a method wherein the first plurality of labeled training examples are labeled with a query intent class ( Paragraph 30, “the training queries include the NER and intent labels, including for at least one query intent label“ ) Shah is considered analogous to the invention because both concern the field of artificial intelligence. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified Yang, Mirzadeh, and Abdi to incorporate the teachings of Shah. Doing so ensures improvements in accuracy and responsiveness for recognizing user query intents ( see Paragraph 3 of Shah ). Claim 19 is a system claim corresponding to method claim 10 and is rejected for the same reasons as given in the rejection of that claim. Conclusion Any inquiry concerning this communication or earlier communications from the examiner should be directed to MATTHEWOS MESFIN whose telephone number is (571)270-0782. The examiner can normally be reached Monday-Friday 8am-5pm. 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, Cesar Paula, can be reached at (571) 272-4128. 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. /MATTHEWOS MESFIN/Examiner, Art Unit 2145 /CESAR B PAULA/Supervisory Patent Examiner, Art Unit 2145 Application/Control Number: 18/491,877 Page 2 Art Unit: 2145 Application/Control Number: 18/491,877 Page 3 Art Unit: 2145 Application/Control Number: 18/491,877 Page 4 Art Unit: 2145 Application/Control Number: 18/491,877 Page 5 Art Unit: 2145 Application/Control Number: 18/491,877 Page 6 Art Unit: 2145 Application/Control Number: 18/491,877 Page 7 Art Unit: 2145 Application/Control Number: 18/491,877 Page 8 Art Unit: 2145 Application/Control Number: 18/491,877 Page 9 Art Unit: 2145 Application/Control Number: 18/491,877 Page 10 Art Unit: 2145 Application/Control Number: 18/491,877 Page 11 Art Unit: 2145 Application/Control Number: 18/491,877 Page 12 Art Unit: 2145 Application/Control Number: 18/491,877 Page 13 Art Unit: 2145 Application/Control Number: 18/491,877 Page 14 Art Unit: 2145 Application/Control Number: 18/491,877 Page 15 Art Unit: 2145 Application/Control Number: 18/491,877 Page 16 Art Unit: 2145 1 The large corpus being the second plurality (the first is the base corpus) 2 The claim language includes “query processing model” and “for the second plurality of unlabeled training examples”, which isn’t taught by Mirzadeh. Those limitations are present in Yang ( re page 8) and can be combined to cover the statement as a whole. 3 The claim language includes “unlabeled” in regards to the training examples, which isn’t taught by Abdi. Those limitations are present in Yang, ( re page 8 ), and can be combined to cover the statement as a whole. 4 The claim language includes “query processing model”, which isn’t taught by Mirzadeh. Those limitations are present in Yang, as shown above. 5 The claim language includes “query processing model”, which isn’t taught by Mirzadeh. Those limitations are present in Yang, as shown above. 6 “This number is roughly proportional to the actual size or number of parameters of the neural network” Mirzadeh, Page 5193 7 The claim language includes “query processing model” and “queries”, which isn’t taught by Mou. Those limitations are present in Yang ( re page 8) and can be combined to cover the statement as a whole.
Prosecution Insights