Automatic Extraction of Commonsense LocatedNear Knowledge

TL;DR

This paper presents a method for automatically extracting commonsense 'LocatedNear' relationships between objects from text using a sentence-level classifier and score aggregation, along with new benchmark datasets.

Contribution

It introduces a novel approach combining sentence-level classification with corpus-wide score aggregation and provides benchmark datasets for the task.

Findings

Effective extraction of 'LocatedNear' relations demonstrated.

Benchmark datasets released for future research.

Method shows promise for improving commonsense knowledge extraction.

Abstract

LocatedNear relation is a kind of commonsense knowledge describing two physical objects that are typically found near each other in real life. In this paper, we study how to automatically extract such relationship through a sentence-level relation classifier and aggregating the scores of entity pairs from a large corpus. Also, we release two benchmark datasets for evaluation and future research.