[1Win4] Poster session 1

[1Win4-32] Improvement of Business Segement-level Sentiment Analysis on Financial Documents using Expanded Business Segment Related Terms

Tomoki Ito¹, 〇Ryuichi Tachibana², Chiaki Murata³, Jun Takayama³ (1.Mitsui & Co. Ltd., 2.MITSUI KNOWLEDGE INDUSTRY CO., LTD., 3.Tohoku University)

Keywords:Medical Natural Language Processing, Artificial Intelligence

There is a great demand for the usage of Real World Data (RWD) for medical domains because of the remarkable progress in large language models (LLMs). One potential application of such diverse medical data should be in the pharmaceutical and clinical trial domains. In such a situation, evaluation benchmark for this domain is required; however, this type of benchmarks and evaluation datasets are not sufficiently prepared. Therefore, in this paper we develop an evaluation dataset for information extraction related to medication and clinical trials. Moreover, we evaluate several LLMs using the developed dataset, and investigate the property of each LLM.