2023-05-11 更新
CodeIE: Large Code Generation Models are Better Few-Shot Information Extractors
Authors:Peng Li, Tianxiang Sun, Qiong Tang, Hang Yan, Yuanbin Wu, Xuanjing Huang, Xipeng Qiu
Large language models (LLMs) pre-trained on massive corpora have demonstrated impressive few-shot learning ability on many NLP tasks. A common practice is to recast the task into a text-to-text format such that generative LLMs of natural language (NL-LLMs) like GPT-3 can be prompted to solve it. However, it is nontrivial to perform information extraction (IE) tasks with NL-LLMs since the output of the IE task is usually structured and therefore is hard to be converted into plain text. In this paper, we propose to recast the structured output in the form of code instead of natural language and utilize generative LLMs of code (Code-LLMs) such as Codex to perform IE tasks, in particular, named entity recognition and relation extraction. In contrast to NL-LLMs, we show that Code-LLMs can be well-aligned with these IE tasks by designing code-style prompts and formulating these IE tasks as code generation tasks. Experiment results on seven benchmarks show that our method consistently outperforms fine-tuning moderate-size pre-trained models specially designed for IE tasks (e.g., UIE) and prompting NL-LLMs under few-shot settings. We further conduct a series of in-depth analyses to demonstrate the merits of leveraging Code-LLMs for IE tasks.
PDF Accepted to ACL 2023 (main conference). Code and data are publicly available at https://github.com/dasepli/CodeIE
点此查看论文截图
Are ChatGPT and GPT-4 General-Purpose Solvers for Financial Text Analytics? An Examination on Several Typical Tasks
Authors:Xianzhi Li, Xiaodan Zhu, Zhiqiang Ma, Xiaomo Liu, Sameena Shah
The most recent large language models such as ChatGPT and GPT-4 have garnered significant attention, as they are capable of generating high-quality responses to human input. Despite the extensive testing of ChatGPT and GPT-4 on generic text corpora, showcasing their impressive capabilities, a study focusing on financial corpora has not been conducted. In this study, we aim to bridge this gap by examining the potential of ChatGPT and GPT-4 as a solver for typical financial text analytic problems in the zero-shot or few-shot setting. Specifically, we assess their capabilities on four representative tasks over five distinct financial textual datasets. The preliminary study shows that ChatGPT and GPT-4 struggle on tasks such as financial named entity recognition (NER) and sentiment analysis, where domain-specific knowledge is required, while they excel in numerical reasoning tasks. We report both the strengths and limitations of the current versions of ChatGPT and GPT-4, comparing them to the state-of-the-art finetuned models as well as pretrained domain-specific generative models. Our experiments provide qualitative studies, through which we hope to help understand the capability of the existing models and facilitate further improvements.
PDF 9 pages, 5 figures
点此查看论文截图
Few-shot Link Prediction on N-ary Facts
Authors:Jiyao Wei, Saiping Guan, Xiaolong Jin, Jiafeng Guo, Xueqi Cheng
N-ary facts composed of a primary triple (head entity, relation, tail entity) and an arbitrary number of auxiliary attribute-value pairs, are prevalent in real-world knowledge graphs (KGs). Link prediction on n-ary facts is to predict a missing element in an n-ary fact. This helps populate and enrich KGs and further promotes numerous downstream applications. Previous studies usually require a substantial amount of high-quality data to understand the elements in n-ary facts. However, these studies overlook few-shot relations, which have limited labeled instances, yet are common in real-world scenarios. Thus, this paper introduces a new task, few-shot link prediction on n-ary facts. It aims to predict a missing entity in an n-ary fact with limited labeled instances. We further propose a model for Few-shot Link prEdict on N-ary facts, thus called FLEN, which consists of three modules: the relation learning, support-specific adjusting, and query inference modules. FLEN captures relation meta information from limited instances to predict a missing entity in a query instance. To validate the effectiveness of FLEN, we construct three datasets based on existing benchmark data. Our experimental results show that FLEN significantly outperforms existing related models in both few-shot link prediction on n-ary facts and binary facts.
PDF 12 pages, submitted to IEEE for possible publication
点此查看论文截图