Iterative Document-level Information Extraction via Imitation Learning

TL;DR

This paper introduces IterX, an imitation learning-based iterative model for document-level relation extraction that effectively identifies and extracts complex templates without predefined order, achieving state-of-the-art results.

Contribution

The paper proposes a novel imitation learning approach for iterative document-level extraction, removing the need for predefined template sequences and improving extraction performance.

Findings

Achieves state-of-the-art on SciREX and MUC-4 benchmarks.

Effective in extracting complex relation templates.

Performs well on a new granular extraction task.

Abstract

We present a novel iterative extraction model, IterX, for extracting complex relations, or templates (i.e., N-tuples representing a mapping from named slots to spans of text) within a document. Documents may feature zero or more instances of a template of any given type, and the task of template extraction entails identifying the templates in a document and extracting each template's slot values. Our imitation learning approach casts the problem as a Markov decision process (MDP), and relieves the need to use predefined template orders to train an extractor. It leads to state-of-the-art results on two established benchmarks -- 4-ary relation extraction on SciREX and template extraction on MUC-4 -- as well as a strong baseline on the new BETTER Granular task.