Named Entity Recognition Based on Large Language Model and Instruction Tuning
Sheng Yang, Min Li, Xu Luo · 2024
In recent years, with the development of deep learning technology, Large Language models (LLMs) perform well on many natural language processing tasks, but the performance on named entity recognition tasks are average. This is because of the difference between named entity recognition and Large Language Model: the former is essentially a sequence labeling task, while the latter is a text generation model. Therefore, in this paper, we propose a named entity recognition method based on large language model and instruction fine-tuning. First, combining the idea of prompt engineering, we reconstruct the data set format of named entities by constructing instruction prompt to fit the generation task. The instruction cue consists of four parts: task description, entity definition, task example and input text. In addition, this paper also adopts the method of instruction fine-tuning to further improve the capability of named entity recognition for large language models. Finally, this paper selects the CLUENER2020 Chinese named entity recognition dataset for experimental validation, and compares with several traditional named entity recognition methods. The results show that the method proposed in this paper performs better.