Stretching Sentence-pair NLI Models to Reason over Long Documents and Clusters

TL;DR

This paper extends NLI models to handle long documents and clusters, improving document-level reasoning and evidence retrieval, and revealing inconsistencies across sources.

Contribution

It introduces new aggregation methods for NLI models to operate on full documents and clusters, achieving state-of-the-art results and uncovering source discrepancies.

Findings

State-of-the-art performance on ContractNLI dataset

NLI scores provide strong retrieval signals

Detection of inconsistencies across multilingual Wikipedia pages

Abstract

Natural Language Inference (NLI) has been extensively studied by the NLP community as a framework for estimating the semantic relation between sentence pairs. While early work identified certain biases in NLI models, recent advancements in modeling and datasets demonstrated promising performance. In this work, we further explore the direct zero-shot applicability of NLI models to real applications, beyond the sentence-pair setting they were trained on. First, we analyze the robustness of these models to longer and out-of-domain inputs. Then, we develop new aggregation methods to allow operating over full documents, reaching state-of-the-art performance on the ContractNLI dataset. Interestingly, we find NLI scores to provide strong retrieval signals, leading to more relevant evidence extractions compared to common similarity-based methods. Finally, we go further and investigate whole document clusters to identify both discrepancies and consensus among sources. In a test case, we find real inconsistencies between Wikipedia pages in different languages about the same topic.