MUMBAI, India, July 13 -- Intellectual Property India has published a patent application (202441002887 A) filed by Cognizant Technology Solutions India Pvt. Ltd. on January 15, 2024, for Gen Ai-Based System And Method For Iterative Refinement Of Innovation Data Using Quality Score.
Inventors include Alexis Samuel; Pandiyan Adiyapatham; Balamurugan Padmanaban; Rengaraj H; Sajal Garg; and Ananthi E.
The application for the patent was published on July 10, 2026, under issue no. 28/2026.
Abstract: We Claim: 1. A system (100) for iterative refinement of innovation data, the system comprising: 5 a memory (120) storing program instructions; a processor (118) executing program instructions stored in the memory (120) and configured to execute an innovation data refinement engine (124) to: fetch input data and innovation data to validate the 10 input data and the innovation data to generate a validated data, wherein the innovation data represents data related to an innovation process of a project development lifecycle; determine a quality score for the validated data 15 based on a weighted score of one or more predefined parameters; generate a prompt data from the validated data using employing an NLP model; enhance the prompt data based on the quality score 20 employing enhancement rules to generate an enhanced prompt data; assign an enhanced quality score to the enhanced prompt data to generate a modified prompt data; and generate features representative of refined 25 innovation data for generating a code for deployment based on the modified prompt data. 2. The system (100) as claimed in claim 1, wherein the innovation data refinement engine (124) fetches the input 30 data from an input data unit 104, the input data includes forecasted values based on one or more benefits foreseen upon implementation of an idea, pre-defined workflow, idea descriptions and problem statements, chat messages, an opportunity type, description of business problems, current 35 scenarios, one or more issues faced by end-users, IT teams, 25 clients, data related to hackathons, and crowdsourced ideas where multiple users share ideas. 3. The system (100) as claimed in claim 1, wherein the 5 innovation data refinement engine (124) fetches the innovation data from an innovation data unit 102, the innovation data includes data relating to historical projects and engagements related data, data related to past and ongoing innovation initiatives, and project details, 10 outcomes, challenges, and other relevant information. 4. The system (100) as claimed in claim 1, wherein the innovation data refinement engine (124) comprises a validation unit (108) configured to validate the input data 15 and the innovation data based on one or more pre-defined grammar rules employing the NLP model in an NLP unit (110) of the innovation data refinement engine (124). 5. The system (100) as claimed in claim 4, wherein the system 20 (100) comprises a user interface 106 for rendering previous ideas entered by one or more users against an identified opportunity, and wherein the validation unit (108) accesses the tagged input data and innovation data in the reusable asset for carrying out the validation. 25 6. The system (100) as claimed in claim 4, wherein the validation unit (108) is configured to validate the input data and innovation based on pre-defined grammar rules by employing the NLP model stored in the NLP unit (110) of the 30 innovation data refinement engine, and wherein the validation unit is configured to: block one or more changes to a structure of the input data and innovation data; block unsafe Hypertext Markup Language (HTML) or 35 JavaScript content in the input data and the innovation data; 26 block the input data and the innovation data that attempt to access or modify system data or configuration; block the input data and the innovation data that attempt to modify the given system prompt and other harmful 5 data in the input data and innovation data; block the input data and the innovation data which are unrelated to technology and language standards; prohibit jailbreak of large language models; and prohibit malicious code, uniform resource locator, 10 link or website cyber security threat in the input data and the innovation data. 7. The system as claimed in claim 1, wherein the innovation data refinement engine (124) comprises a score generation 15 unit (112) configured to determine the quality score by dividing the total weighted score of the predefined parameters by a total number of second nested parameter and first nested parameters for which no second nested parameters exist. 20 8. The system (100) as claimed in claim 7, wherein the one or more predefined parameters represent one or more attributes for assessing potential and impact of innovation, and wherein the predefined parameters are associated with 25 one or more first nested parameters, the first nested parameters represent one or more specific characteristics associated with the pre-defined parameters, and wherein the one or more first nested parameters are associated with one or more second nested parameters, the second nested 30 parameters represent one or more specific categories associated with the first nested parameter. 9. The system (100) as claimed in claim 7, wherein the score generation unit (112) assigns the quality score to the 35 validated data based on an evaluation matrix, wherein the evaluation matrix is obtained based on values obtained for 27 the predefined parameters, the first nested parameters and the second nested parameters. 10. The system (100) as claimed in claim 1, wherein the 5 innovation data refinement engine (124) comprises a data enhancement unit (114) configured to: generate the prompt data associated with the validation data employing the NLP model stored in an NLP unit (110) of 10 the innovation data refinement engine (124): generate the enhanced prompt data based on the quality score employing the enhancement rules to generate the enhanced prompt data, wherein the enhancement rules are generated basis a determination of a context of the 15 validated data in terms of one or more features including, problem-statement, title, idea description, and enhancing the prompt data by reconstructing the prompt data by classifying the validated data in terms of one or more enhancement parameters including persona, task, input 20 elements, generative, directive along with the context; and enhance the validated data based on the enhanced prompt data employing one or more additional inputs using the NLP model. 25 11. The system (100) as claimed in claim 10, wherein the innovation data refinement engine (124) comprises a score generation unit (112) configured to: assign the enhanced quality score to the enhanced prompt data received from the data enhancement unit (114); 30 transmit the enhanced prompt data with the assigned enhanced quality score to the data enhancement unit (114); and enable generation of the modified prompt data based on the enhanced quality score. 35 28 12. The system (100) as claimed in claim 11, wherein the innovation data refinement engine (124) comprises a story generation unit (116) configured to generate the features based on the modified prompt data employing LLMs, wherein 5 the features represent user story data and epic data. 13. The system (100) as claimed in claim 12, wherein the innovation data refinement engine (124) comprises an output unit (122) configured to generate the code for deployment 10 based on the generated features. 14. A method for iterative refinement of innovation data, the method comprising steps of: fetching input data and innovation data to validate 15 the input data and the innovation data to generate a validated data, wherein the innovation data represents data related to an innovation process of a project development lifecycle; determining a quality score for the validated data 20 based on a weighted score of one or more predefined parameters; generating a prompt data from the validated data employing an NLP model; enhancing the prompt data based on the quality score 25 employing enhancement rules to generate an enhanced prompt data; assigning an enhanced quality score to the enhanced prompt data to generate a modified prompt data; and generating features representative of refined 30 innovation data for generating a code for deployment based on the modified prompt data. 15. The method as claimed in claim 14, wherein the input data includes forecasted values based on one or more 35 benefits foreseen upon implementation of an idea, pre- 29 defined workflow, idea descriptions and problem statements, chat messages, an opportunity type, description of business problems, current scenarios, one or more issues faced by end-users, IT teams, clients, data related to hackathons, 5 and crowdsourced ideas where multiple users share ideas. 16. The method as claimed in claim 14, wherein the innovation data includes data relating to historical projects and engagements related data, data related to past 10 and ongoing innovation initiatives, and project details, outcomes, challenges, and other relevant information. 17. The method as claimed in claim 14, wherein the step of validating the input data and the innovation data comprises 15 validating the input data and innovation based on predefined grammar rules by employing the NLP model, and wherein the step of validating the input data and the innovation data comprises the steps of: blocking one or more changes to a structure of the 20 input data and innovation data; blocking unsafe Hypertext Markup Language (HTML) or JavaScript content in the input data and the innovation data; blocking the input data and the innovation data that 25 attempt to access or modify system data or configuration; blocking the input data and the innovation data that attempt to modify the given system prompt and other harmful data in the input data and the innovation data; blocking the input data and the innovation data which 30 are unrelated to technology and language standards; prohibiting jailbreak of large language models; and prohibiting malicious code, Uniform Resource Locator (URL), link or website cyber security threat in the input data and the innovation data. 35 30 18. The method as claimed in claim 14, wherein the step of determining the quality score for the validated data comprises: determining the quality score by dividing the total 5 weighted score of the predefined parameters by a total number of second nested parameter and first nested parameters for which no second nested parameters exist , wherein the quality score is assigned to the validated data based on an evaluation matrix, the evaluation matrix is 10 obtained based on values obtained for the predefined parameters, the first nested parameters and the second nested parameters. 19. The method as claimed in claim 18, wherein the one or 15 more predefined parameters represent one or more attributes for assessing potential and impact of innovation, and wherein the predefined parameters are associated with one or more first nested parameters, the first nested parameters represent one or more specific characteristics associated 20 with the pre-defined parameters, and wherein the one or more first nested parameters are associated with one or more second nested parameters, the second nested parameters represent one or more specific categories associated with the first nested parameter. 25 20. The method as claimed in claim 14, wherein the step of enhancing the prompt data to generate the enhanced prompt data comprises the steps of: 30 generating the enhanced prompt data employing the enhancement rules based on the quality score, wherein the enhancement rules are generated basis a determination of a context of the validated data in terms of one or more features including, problem-statement, title, idea 35 description, and enhancing the prompt data by reconstructing the prompt data by classifying the validated data in terms 31 of one or more enhancement parameters including persona, task, input elements, generative, directive along with the context; and enhancing the validated data based on the enhanced 5 prompt data employing one or more additional inputs using the NLP model.
Disclaimer: Curated by HT Syndication.
