ASPER: Attention-based Approach to Extract Syntactic Patterns denoting Semantic Relations in Sentential Context

TL;DR

ASPER is an attention-based deep learning model that automatically extracts syntactic patterns indicating semantic relations between entities in sentences, improving over existing methods across multiple datasets.

Contribution

The paper introduces ASPER, a novel supervised deep learning approach utilizing attention mechanisms for extracting semantic relation patterns in sentential context.

Findings

ASPER outperforms existing methods in extracting syntactic patterns.

Effective across three semantic relations: hyponym-hypernym, cause-effect, meronym-holonym.

Validated on six diverse datasets.

Abstract

Semantic relationships, such as hyponym-hypernym, cause-effect, meronym-holonym etc. between a pair of entities in a sentence are usually reflected through syntactic patterns. Automatic extraction of such patterns benefits several downstream tasks, including, entity extraction, ontology building, and question answering. Unfortunately, automatic extraction of such patterns has not yet received much attention from NLP and information retrieval researchers. In this work, we propose an attention-based supervised deep learning model, ASPER, which extracts syntactic patterns between entities exhibiting a given semantic relation in the sentential context. We validate the performance of ASPER on three distinct semantic relations -- hyponym-hypernym, cause-effect, and meronym-holonym on six datasets. Experimental results show that for all these semantic relations, ASPER can automatically identify a collection of syntactic patterns reflecting the existence of such a relation between a pair of entities in a sentence. In comparison to the existing methodologies of syntactic pattern extraction, ASPER's performance is substantially superior.