Definition Frames: Using Definitions for Hybrid Concept Representations

TL;DR

This paper introduces Definition Frames, a new interpretable word representation method derived from definitions, which captures semantic relations and performs competitively on word similarity tasks.

Contribution

The paper proposes Definition Frames, a novel semantic representation based on definitions that enhances interpretability while maintaining competitive performance.

Findings

Definition Frames are semantically interpretable.

DFs perform well on word similarity tasks.

DFs relate to Qualia structure relations.

Abstract

Advances in word representations have shown tremendous improvements in downstream NLP tasks, but lack semantic interpretability. In this paper, we introduce Definition Frames (DF), a matrix distributed representation extracted from definitions, where each dimension is semantically interpretable. DF dimensions correspond to the Qualia structure relations: a set of relations that uniquely define a term. Our results show that DFs have competitive performance with other distributional semantic approaches on word similarity tasks.