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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Structural data well exists in Web applications, such as social networks insocial media, citation networks in academic websites, and threads data inonline forums. Due to the complex topology, it is difficult to process and makeuse of the rich information within such data. Graph Neural Networks (GNNs) haveshown great advantages on learning representations for structural data.However, the non-transparency of the deep learning models makes it non-trivialto explain and interpret the predictions made by GNNs. Meanwhile, it is also abig challenge to evaluate the GNN explanations, since in many cases, theground-truth explanations are unavailable.  In this paper, we take insights of Counterfactual and Factual (CF^2)reasoning from causal inference theory, to solve both the learning andevaluation problems in explainable GNNs. For generating explanations, wepropose a model-agnostic framework by formulating an optimization problem basedon both of the two casual perspectives. This distinguishes CF^2 from previousexplainable GNNs that only consider one of them. Another contribution of thework is the evaluation of GNN explanations. For quantitatively evaluating thegenerated explanations without the requirement of ground-truth, we designmetrics based on Counterfactual and Factual reasoning to evaluate the necessityand sufficiency of the explanations. Experiments show that no matterground-truth explanations are available or not, CF^2 generates betterexplanations than previous state-of-the-art methods on real-world datasets.Moreover, the statistic analysis justifies the correlation between theperformance on ground-truth evaluation and our proposed metrics. Source code isavailable at https://github.com/chrisjtan/gnn_cff.

Learning and Evaluating Graph Neural Network Explanations based on Counterfactual and Factual Reasoning