Evaluating Document Coherence Modelling

TL;DR

This paper introduces INSteD, a large dataset for evaluating document coherence modeling via sentence intrusion detection, revealing pretrained language models excel in-domain but struggle cross-domain, highlighting room for improvement.

Contribution

It presents INSteD, a new large-scale dataset for sentence intrusion detection, and evaluates pretrained language models' discourse modeling capabilities across domains.

Findings

Pretrained LMs perform well in in-domain settings.

Performance drops significantly in cross-domain evaluations.

There is substantial room for improvement in discourse modeling.

Abstract

While pretrained language models ("LM") have driven impressive gains over morpho-syntactic and semantic tasks, their ability to model discourse and pragmatic phenomena is less clear. As a step towards a better understanding of their discourse modelling capabilities, we propose a sentence intrusion detection task. We examine the performance of a broad range of pretrained LMs on this detection task for English. Lacking a dataset for the task, we introduce INSteD, a novel intruder sentence detection dataset, containing 170,000+ documents constructed from English Wikipedia and CNN news articles. Our experiments show that pretrained LMs perform impressively in in-domain evaluation, but experience a substantial drop in the cross-domain setting, indicating limited generalisation capacity. Further results over a novel linguistic probe dataset show that there is substantial room for improvement, especially in the cross-domain setting.