Reading the Manual: Event Extraction as Definition Comprehension

TL;DR

This paper explores whether event extraction can be achieved through understanding and refining declarative definitions from annotation manuals, enabling end-users to easily construct and extend extraction frameworks.

Contribution

It introduces a model that uses declarative statements for event extraction, demonstrating the ability to generalize to unseen event types by reading new definitions.

Findings

Effective extraction of events under closed ontologies

Generalization to unseen event types by reading new definitions

Model employs incremental refinement of declarative statements

Abstract

We ask whether text understanding has progressed to where we may extract event information through incremental refinement of bleached statements derived from annotation manuals. Such a capability would allow for the trivial construction and extension of an extraction framework by intended end-users through declarations such as, "Some person was born in some location at some time." We introduce an example of a model that employs such statements, with experiments illustrating we can extract events under closed ontologies and generalize to unseen event types simply by reading new definitions.