Unsupervised Semantic Frame Induction using Triclustering

TL;DR

This paper introduces Triframes, a graph-based method for unsupervised semantic frame induction using triclustering of dependency triples, achieving state-of-the-art results on FrameNet data.

Contribution

It presents a novel triclustering approach for semantic frame induction from dependency triples, generalizing clustering for triadic data.

Findings

State-of-the-art performance on FrameNet dataset

Competitive results on verb class clustering

Effective unsupervised semantic frame induction

Abstract

We use dependency triples automatically extracted from a Web-scale corpus to perform unsupervised semantic frame induction. We cast the frame induction problem as a triclustering problem that is a generalization of clustering for triadic data. Our replicable benchmarks demonstrate that the proposed graph-based approach, Triframes, shows state-of-the art results on this task on a FrameNet-derived dataset and performing on par with competitive methods on a verb class clustering task.