Abstract
The present disclosure relates to a method and system for enhancing the accuracy and flexibility of Large Language Models (LLMs) during complex question answering. The method involves identifying a plurality of decision nodes within the LLM's internal reasoning structure, each representing a point of ambiguity or multiple potential reasoning paths. The reasoning process is paused at each identified decision node to present contextually relevant options to the user. User input is received at these decision nodes, specifying preferences or additional data to guide subsequent reasoning. This input is integrated into the ongoing reasoning process of the LLM, dynamically generating a response that reflects improved contextual relevance and aligns with user specifications. Furthermore, the system comprises a decision node identification module, a user input interface, a reasoning adaptation module, and a dynamic response generation module to facilitate enhanced collaborative problem-solving during complex inquiries.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Sinha, Ravi Shanker Kumar; Mohan, Shivam; Raj, Satyam; and Dhamija, Dheeraj, "METHOD AND APPROACH IN LARGE LANGUAGE MODELS FOR COMPLEX QUESTION ANSWERING", Technical Disclosure Commons, (January 09, 2026)
https://www.tdcommons.org/dpubs_series/9161