English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The opaqueness of the multi-hop fact verification model imposes imperativerequirements for explainability. One feasible way is to extract rationales, asubset of inputs, where the performance of prediction drops dramatically whenbeing removed. Though being explainable, most rationale extraction methods formulti-hop fact verification explore the semantic information within each pieceof evidence individually, while ignoring the topological informationinteraction among different pieces of evidence. Intuitively, a faithfulrationale bears complementary information being able to extract otherrationales through the multi-hop reasoning process. To tackle suchdisadvantages, we cast explainable multi-hop fact verification as subgraphextraction, which can be solved based on graph convolutional network (GCN) withsalience-aware graph learning. In specific, GCN is utilized to incorporate thetopological interaction information among multiple pieces of evidence forlearning evidence representation. Meanwhile, to alleviate the influence ofnoisy evidence, the salience-aware graph perturbation is induced into themessage passing of GCN. Moreover, the multi-task model with three diagnosticproperties of rationale is elaborately designed to improve the quality of anexplanation without any explicit annotations. Experimental results on theFEVEROUS benchmark show significant gains over previous state-of-the-artmethods for both rationale extraction and fact verification.

Exploring Faithful Rationale for Multi-hop Fact Verification via Salience-Aware Graph Learning