Applying the Ego Network Model to Cross-Target Stance Detection

TL;DR

This paper introduces the Ego Network Model and Signed Ego Network Model as novel, data-efficient tools for cross-target stance detection, achieving comparable accuracy to graph-based methods by leveraging social network features.

Contribution

The paper presents two new social network-based models for stance detection that require less data and outperform traditional text-only approaches.

Findings

Outer social ties are more predictive of stance than inner ties.

ENM and SENM achieve similar accuracy to graph-based models like CT-TN.

Models require less data and are easier to obtain.

Abstract

Understanding human interactions and social structures is an incredibly important task, especially in such an interconnected world. One task that facilitates this is Stance Detection, which predicts the opinion or attitude of a text towards a target entity. Traditionally, this has often been done mainly via the use of text-based approaches, however, recent work has produced a model (CT-TN) that leverages information about a user's social network to help predict their stance, outperforming certain cross-target text-based approaches. Unfortunately, the data required for such graph-based approaches is not always available. This paper proposes two novel tools for Stance Detection: the Ego Network Model (ENM) and the Signed Ego Network Model (SENM). These models are founded in anthropological and psychological studies and have been used within the context of social network analysis and related tasks (e.g., link prediction). Stance Detection predictions obtained using these features achieve a level of accuracy similar to the graph-based features used by CT-TN while requiring less and more easily obtainable data. In addition to this, the performances of the inner and outer circles of the ENM, representing stronger and weaker social ties, respectively are compared. Surprisingly, the outer circles, which contain more numerous but less intimate connections, are more useful for predicting stance.