Joint Constrained Learning for Event-Event Relation Extraction

TL;DR

This paper introduces a joint constrained learning framework that models event-event relations in natural language, effectively handling data scarcity and outperforming existing methods in temporal and hierarchical event relation tasks.

Contribution

The paper proposes a novel joint constrained learning approach that encodes logical constraints into differentiable objectives, improving event relation extraction without the need for extensive labeled data.

Findings

Outperforms state-of-the-art methods on temporal relation benchmarks

Effectively induces event complexes in external corpora

Replaces expensive global inference with a more efficient approach

Abstract

Understanding natural language involves recognizing how multiple event mentions structurally and temporally interact with each other. In this process, one can induce event complexes that organize multi-granular events with temporal order and membership relations interweaving among them. Due to the lack of jointly labeled data for these relational phenomena and the restriction on the structures they articulate, we propose a joint constrained learning framework for modeling event-event relations. Specifically, the framework enforces logical constraints within and across multiple temporal and subevent relations by converting these constraints into differentiable learning objectives. We show that our joint constrained learning approach effectively compensates for the lack of jointly labeled data, and outperforms SOTA methods on benchmarks for both temporal relation extraction and event hierarchy construction, replacing a commonly used but more expensive global inference process. We also present a promising case study showing the effectiveness of our approach in inducing event complexes on an external corpus.