Language Models for Lexical Inference in Context

TL;DR

This paper explores the use of pretrained language models for lexical inference in context, introducing three novel approaches that outperform previous methods in recognizing entailment between similar sentences with lexical differences.

Contribution

It presents the first pretrained LM-based methods for LIiC, including a few-shot classifier and pattern-based relation induction approaches, demonstrating their effectiveness.

Findings

All proposed methods outperform previous state-of-the-art.

Pretrained LMs show strong potential for lexical inference tasks.

Analysis reveals factors influencing success and failure of the approaches.

Abstract

Lexical inference in context (LIiC) is the task of recognizing textual entailment between two very similar sentences, i.e., sentences that only differ in one expression. It can therefore be seen as a variant of the natural language inference task that is focused on lexical semantics. We formulate and evaluate the first approaches based on pretrained language models (LMs) for this task: (i) a few-shot NLI classifier, (ii) a relation induction approach based on handcrafted patterns expressing the semantics of lexical inference, and (iii) a variant of (ii) with patterns that were automatically extracted from a corpus. All our approaches outperform the previous state of the art, showing the potential of pretrained LMs for LIiC. In an extensive analysis, we investigate factors of success and failure of our three approaches.