Conversational Financial Information Retrieval Model (ConFIRM)

  • 2023-11-10 07:15:17
  • Stephen Choi, William Gazeley, Siu Ho Wong, Tingting Li
  • 0


With the exponential growth in large language models (LLMs), leveraging theiremergent properties for specialized domains like finance merits exploration.However, regulated fields such as finance pose unique constraints, requiringdomain-optimized frameworks. We present ConFIRM, an LLM-based conversationalfinancial information retrieval model tailored for query intent classificationand knowledge base labeling. ConFIRM comprises two modules: 1) a method to synthesize finance domain-specific question-answer pairs, and 2) evaluation of parameter efficient fine-tuning approaches for the queryclassification task. We generate a dataset of over 4000 samples, assessingaccuracy on a separate test set. ConFIRM achieved over 90% accuracy, essential for regulatory compliance.ConFIRM provides a data-efficient solution to extract precise query intent forfinancial dialog systems.


Quick Read (beta)

loading the full paper ...