A Factorized Model for Transitive Verbs in Compositional Distributional Semantics

TL;DR

This paper introduces a novel factorized compositional distributional semantics model for transitive verbs that effectively combines noun and verb representations to improve performance on semantic tasks.

Contribution

It proposes a new factorized approach that constructs (subject, verb) and (verb, object) vectors and combines them for better transitive verb representation.

Findings

Outperforms recent models on established transitive verb tasks

Uses simple vector operations for combining representations

Provides a scalable approach for compositional semantics

Abstract

We present a factorized compositional distributional semantics model for the representation of transitive verb constructions. Our model first produces (subject, verb) and (verb, object) vector representations based on the similarity of the nouns in the construction to each of the nouns in the vocabulary and the tendency of these nouns to take the subject and object roles of the verb. These vectors are then combined into a final (subject,verb,object) representation through simple vector operations. On two established tasks for the transitive verb construction our model outperforms recent previous work.