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 § 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)(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.
Claims 1-7, 10-16, and 19-20 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Michael et al. US 20240320450 A1 (hereinafter Michael).
Regarding independent claims 1, 10, and 19, Michael teaches a method for mixed language text understanding for Generative Artificial Intelligence (GenAI) models, the method comprising / a computing device for mixed language text understanding for Generative Artificial Intelligence (GenAI) models, the computing device comprising / a non-transitory computer-readable medium storing computer-executable instructions for mixed language text understanding for Generative Artificial Intelligence (GenAI) models, the computer-executable instructions configured for:
a processor (FIG. 1, 104); and
a memory communicatively coupled to the processor, wherein the memory stores processor-executable instructions, which when executed by the processor, cause the processor to (FIG. 1, 110)
receiving, by a computing device, a raw parallel corpus of two languages, wherein the raw parallel corpus comprises a plurality of samples of cross-domain parallel text data in the two languages (FIG. 3A, 302 [0055] “At step 302, the system 100 receives the code-mixed utterance (input text) from the user for performing the one or more actions via a chatbot”);
generating, by the computing device, a cross-domain codemix parallel corpus and a first set of linguistic features from the raw parallel corpus using statistical and linguistic techniques ([0069] “To adapt the pre-trained language models to the specific domain of the chatbot, mixed languages domain adaptation training is performed”; [0044] “the obtaining module 206 may be configured to obtain contextual information and contextual representations associated with the received code-mixed utterance… the contextual representations are associated with linguistic properties of the code-mixed utterances”; [0035] “the one or more processors 104 are configured to generate a set of vocabulary probabilities for the fixed dimensional representation associated with each token”);
determining, by the computing device, a complexity of each of the plurality of samples of the cross-domain codemix parallel corpus based on a set of complexity parameters ([0051] “The selection of policies depends on the complexity of the conversational AI system, the available data, and the desired level of customization… the identifying module 212 may be configured to update the one or more policies based on the identified one or more actions and the one or more conversational parameters”; [0004] “where a user utters (types or speaks) words specific to mixed-language or in case where the user keeps switching between two languages where the alphabet is common, methods relying on language identification of the words may produce inaccurate interpretation of the user intent”);
preparing, by the computing device, a curriculum learning dataset from the cross-domain codemix parallel corpus based on the complexity of each of the plurality of samples (FIG. 5, 510, [0067] “At step 510, training arguments are created for optimizing hyperparameters”); and
sequentially fine-tuning, by the computing device, a pre-trained multilingual translation model using each of the plurality of samples in the curriculum learning dataset to obtain a generic pre-trained codemix understanding model ([0053] “the AI-based LTN and the AI-based IDN (i.e., Large Language Model (LLM)-based language translation models) are fine-tuned. Further, fine-tuning the LLM-based language translation models is a technique that involves adapting pre-trained models specifically designed for language translation tasks”).
Regarding claims 2 and 11, Michael teaches all of the limitations of claim 1 and 10, upon which claims 2 and 11 depend.
Additionally, Michael teaches further comprising preprocessing, by the computing device, the cross-domain codemix parallel corpus to obtain a pre-processed cross-domain codemix parallel corpus for each language of the two languages, wherein the pre-processed cross-domain codemix parallel corpus comprises cross-domain codemix text data, corresponding cross-domain text data in the language, the first set of linguistic features, and translation data of the language corresponding to the cross-domain codemix text data ([0066] “At step 508, the training data is pre-processed. In pre-processing the training data, the training data is processed to generate the positional encoding and the word embeddings, such that the AI-based LTN understands the features and the set of target language labels”; [0044] “the contextual representations are associated with linguistic properties of the code-mixed utterances. In an exemplary embodiment of the present disclosure, the linguistic properties of the code-mixed utterances may be syntax, semantics, morphology, pragmatics, and the like”; [0035] “The one or more processors 104 are configured to translate the code-mixed utterances to a translation text in the target language based on the generated set of vocabulary probabilities by using the AI-based LTN”).
Regarding claims 3 and 12, Michael teaches all of the limitations of claim 1 and 10, upon which claim 3 and 12 depend.
Additionally, Michael teaches wherein the set of complexity parameters comprises language switching points, language mix index, and lexical rarity ([0004] “where a user utters (types or speaks) words specific to mixed-language or in case where the user keeps switching between two languages where the alphabet is common, methods relying on language identification of the words may produce inaccurate interpretation of the user intent”).
Regarding claims 4 and 13, Michael teaches all of the limitations of claim 1 and 10, upon which claim 4 and 13 depend.
Additionally, Michael teaches wherein preparing the curriculum learning dataset comprises arranging, by the computing device, the plurality of samples of the cross-domain codemix parallel corpus in an order based on the complexity ([0044] “the obtaining module 206 may be configured to determine the order of each token of the set of tokens in a sequence of the set of tokens based on the fixed-dimensional vector and positional information associated with each token of the set of tokens by using the AI-based LTN”).
Regarding claims 5 and 14, Michael teaches all of the limitations of claim 1 and 10, upon which claim 5 and 14 depend.
Additionally, Michael teaches wherein sequentially fine-tuning the pre-trained multilingual translation model comprises individually fine-tuning, by the computing device, the pre-trained multilingual translation model using each sample of the curriculum learning dataset in an increasing order of complexity ([0053] “This fine-tuning process adapts the pre-trained models to specific language pairs, enabling them to capture the idiosyncrasies and nuances of translation in the target domain”).
Regarding claims 6, 15, and 20 Michael teaches all of the limitations of claim 1, 10, and 19, upon which claim 6, 15, and 20 depend.
Additionally, Michael teaches retrieving, by the computing device, domain specific text data in a first language of the two languages from a domain data source ([0069] “Domain-specific training data comprising labeled examples of user inputs and their corresponding intents is utilized. The data used in training the LTN may be obtained from the books and online documents by the bi-lingual community.”);
translating, by the computing device, the domain specific text data from the first language to a second language of the two languages using a pre-trained translation model ([0053] “This fine-tuning process adapts the pre-trained models to specific language pairs, enabling them to capture the idiosyncrasies and nuances of translation in the target domain”);
upon translating, generating, by the computing device, a domain specific parallel corpus of the two languages ([0069] “Once the language models have been mixed language domain adaptation trained on the domain-specific data, the language models are ready for code-mixed language translation”); and
generating, by the computing device, a domain specific codemix parallel corpus and a second set of linguistic features from the domain specific parallel corpus using statistical and linguistic techniques ([0069] “To adapt the pre-trained language models to the specific domain of the chatbot, mixed languages domain adaptation training is performed”; [0044] “the obtaining module 206 may be configured to obtain contextual information and contextual representations associated with the received code-mixed utterance… the contextual representations are associated with linguistic properties of the code-mixed utterances”; [0035] “the one or more processors 104 are configured to generate a set of vocabulary probabilities for the fixed dimensional representation associated with each token”).
Regarding claims 7 and 16, Michael teaches all of the limitations of claim 6 and 15, upon which claim 7 and 16 depend.
Additionally, Michael teaches further comprising pre-processing, by the computing device, the domain specific codemix parallel corpus to obtain a pre-processed domain specific codemix parallel corpus for each language of the two languages, wherein the pre-processed domain specific codemix parallel corpus comprises domain specific codemix text data, corresponding domain specific text data in the language, the second set of linguistic features, and translation data of the language corresponding to the domain specific codemix text data ([0066] “At step 508, the training data is pre-processed. In pre-processing the training data, the training data is processed to generate the positional encoding and the word embeddings, such that the AI-based LTN understands the features and the set of target language labels”; [0044] “the contextual representations are associated with linguistic properties of the code-mixed utterances. In an exemplary embodiment of the present disclosure, the linguistic properties of the code-mixed utterances may be syntax, semantics, morphology, pragmatics, and the like”; [0035] “The one or more processors 104 are configured to translate the code-mixed utterances to a translation text in the target language based on the generated set of vocabulary probabilities by using the AI-based LTN”).
Allowable Subject Matter
Claims 8-9 and 17-18 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Riesa et al. (US 20190347323 A1) teaches a method for identifying codemixed text includes receiving codemixed text and segmenting the codemixed text into a plurality of tokens. Each token includes at least one character and is delineated from any adjacent tokens by a space. For each token of the codemixed text, the method also includes extracting features from the token and predicting a probability distribution over possible languages for the token using a language identifier model configured to receive the extracted features from the token as feature inputs. The method also includes assigning a language to each token of the codemixed text by executing a greedy search on the probability distribution over the possible languages predicted for each respective token.
Lee (US 20210165974 A1) teaches an artificial intelligence apparatus which inputs first language data into a machine translation model to economically train a natural language understanding model of a second language and obtains second language data corresponding to the first language data to train the natural language understanding model.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ZEESHAN SHAIKH whose telephone number is (703)756-1730. The examiner can normally be reached Monday-Friday 7:30AM-5:00PM.
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.
/ZEESHAN MAHMOOD SHAIKH/Examiner, Art Unit 2658
/RICHEMOND DORVIL/Supervisory Patent Examiner, Art Unit 2658
Prosecution Insights