ABCD Team at SOMD 2024: Software Mention Detection in Scholarly Publications with Large Language Models

This paper outlines the ABCD team’s approach to employing LLMs for three subtasks at SOMD 2024, specifically focusing on Software Mention Detection in Scholarly Publications. The task revolves around scientific articles and information, comprising three subtasks: (1) extracting software entities from a given sentence, (2) extracting relevant entities related to software entities from the task I, and (3) determining the relationship between entities extracted from the previous two tasks. Our objective is to gain valuable insights into fine-tuning LLMs using LoRA. The experimental results showcase that our approach has demonstrated competitive performance across all three tasks, securing Top 1, Top 2, and Top 2 rankings for Subtask I, Subtask II, and Subtask III, respectively. We release our source code in this Github repository( https://github.com/Xphi310302/ABCD-team-NLSP.git ).