Generating Fact Checking Summaries for Web Claims

TL;DR

This paper introduces SUMO, a neural attention-based model that not only assesses the correctness of claims using evidence from web documents but also generates extractive summaries explaining its decisions, improving contextual understanding over prior methods.

Contribution

The paper proposes a hierarchical attention mechanism guided by claims and titles, enhancing evidence extraction and explanation generation for fact checking.

Findings

SUMO outperforms previous models on diverse datasets.

Hierarchical attention improves contextual relevance.

Effective evidence extraction and summarization achieved.

Abstract

We present SUMO, a neural attention-based approach that learns to establish the correctness of textual claims based on evidence in the form of text documents (e.g., news articles or Web documents). SUMO further generates an extractive summary by presenting a diversified set of sentences from the documents that explain its decision on the correctness of the textual claim. Prior approaches to address the problem of fact checking and evidence extraction have relied on simple concatenation of claim and document word embeddings as an input to claim driven attention weight computation. This is done so as to extract salient words and sentences from the documents that help establish the correctness of the claim. However, this design of claim-driven attention does not capture the contextual information in documents properly. We improve on the prior art by using improved claim and title guided hierarchical attention to model effective contextual cues. We show the efficacy of our approach on datasets concerning political, healthcare, and environmental issues.