Learning Open Information Extraction of Implicit Relations from Reading Comprehension Datasets

TL;DR

This paper introduces a method to extract implicit relations from reading comprehension datasets, significantly expanding available training data and improving neural models' ability to identify implicit relations.

Contribution

We developed a parse-based tool to convert reading comprehension datasets into large OpenIE datasets and trained a neural model that outperforms previous methods on implicit relation extraction.

Findings

Created a dataset 35 times larger than previous ones.

Neural model trained on this data surpasses prior methods.

Enhanced extraction of implicit relations from text.

Abstract

The relationship between two entities in a sentence is often implied by word order and common sense, rather than an explicit predicate. For example, it is evident that "Fed chair Powell indicates rate hike" implies (Powell, is a, Fed chair) and (Powell, works for, Fed). These tuples are just as significant as the explicit-predicate tuple (Powell, indicates, rate hike), but have much lower recall under traditional Open Information Extraction (OpenIE) systems. Implicit tuples are our term for this type of extraction where the relation is not present in the input sentence. There is very little OpenIE training data available relative to other NLP tasks and none focused on implicit relations. We develop an open source, parse-based tool for converting large reading comprehension datasets to OpenIE datasets and release a dataset 35x larger than previously available by sentence count. A baseline neural model trained on this data outperforms previous methods on the implicit extraction task.