arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2407.08035cs.CLcs.IR
收藏

FsPONER: 面向领域特定场景的少样本提示优化命名实体识别

FsPONER: Few-shot Prompt Optimization for Named Entity Recognition in Domain-specific Scenarios

Technical University of Munich(慕尼黑技术大学);Siemens AG(西门子股份公司)

Yongjian Tang, Rakebul Hasan, Thomas Runkler

AI总结:

本文提出FsPONER方法,通过三种少样本选择策略优化提示,针对工业制造和维护领域,在数据稀缺场景下实现比微调BERT模型更高的F1分数。

摘要:

大型语言模型(LLMs)为命名实体识别(NER)任务提供了新路径。与微调相比,基于LLMs的提示方法避免了训练需求,节省了大量计算资源,并依赖极少的标注数据。先前研究在通用NER基准上实现了与全监督BERT微调方法相当的性能,但未探讨LLMs在领域特定场景下的少样本学习效率。为此,本文引入FsPONER,一种优化少样本提示的新方法,并在工业制造和维护领域评估其性能,使用GPT-4-32K、GPT-3.5-Turbo、LLaMA 2-chat和Vicuna等多模型。FsPONER包含基于随机采样、TF-IDF向量和两者结合的三种少样本选择方法。我们比较了这些方法与通用GPT-NER方法在少样本示例增加时的性能,并评估其最优NER性能与微调BERT和LLaMA 2-chat的对比。在考虑的数据稀缺现实场景中,FsPONER结合TF-IDF在F1分数上比微调模型高出约10%。

英文摘要:

Large Language Models (LLMs) have provided a new pathway for Named Entity Recognition (NER) tasks. Compared with fine-tuning, LLM-powered prompting methods avoid the need for training, conserve substantial computational resources, and rely on minimal annotated data. Previous studies have achieved comparable performance to fully supervised BERT-based fine-tuning approaches on general NER benchmarks. However, none of the previous approaches has investigated the efficiency of LLM-based few-shot learning in domain-specific scenarios. To address this gap, we introduce FsPONER, a novel approach for optimizing few-shot prompts, and evaluate its performance on domain-specific NER datasets, with a focus on industrial manufacturing and maintenance, while using multiple LLMs -- GPT-4-32K, GPT-3.5-Turbo, LLaMA 2-chat, and Vicuna. FsPONER consists of three few-shot selection methods based on random sampling, TF-IDF vectors, and a combination of both. We compare these methods with a general-purpose GPT-NER method as the number of few-shot examples increases and evaluate their optimal NER performance against fine-tuned BERT and LLaMA 2-chat. In the considered real-world scenarios with data scarcity, FsPONER with TF-IDF surpasses fine-tuned models by approximately 10% in F1 score.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑