CodeIE: Large Code Generation Models are Better Few-Shot Information Extractors

TL;DR

This work tackles the problem of few-shot information extraction for tasks like NER and RE, where outputs are structured rather than plain text. It introduces CodeIE, a method that reformulates IE outputs as Python code and leverages Code-LLMs to generate structured predictions from code-style prompts, including in-context demonstrations. Empirical results on seven benchmarks show CodeIE outperforms UIE fine-tuning and NL-LLMs in few-shot settings, with Code-LLMs delivering higher fidelity and better structural alignment. Analyses reveal that the code-like representation aligns with pretraining on code and reduces structural errors, suggesting Code-LLMs can offer robust, data-efficient IE capabilities. The approach highlights the potential of leveraging code pretraining to improve structured prediction tasks in NLP.

Abstract

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.

Figures (1)

• Figure 1: Illustrations of performing structured NER task with NL-LLMs and Code-LLMs, respectively. In contrast to prompting NL-LLMs with plain natural language, we utilize Code-LLMs with structured code-style prompts to mitigate the output discrepancy between the pre-training and inference stages.