Make Social Networks Clean Again: Graph Embedding and Stacking Classifiers for Bot Detection

Make Social Networks Clean Again: Graph Embedding and Stacking Classifiers for Bot Detection

Авторы

K. Skorniakov, D. Turdakov, and A. Zhabotinsky

Аннотация

The paper introduces a novel approach to the detection of social bots using ensembling of classifiers. We also studied the impact of different feature sets and demonstrated the power of graph em- bedding which is underused by the existing methods. The main contribution of this work is a creating of a stacking based ensem- ble, which effectively exploits text and graph features. Empirical evaluation proved the effectiveness of the proposed method for bots detection and showed improvement in comparison to existing solutions by 4-9 points of AUC.

Полный текст статьи в формате pdf (на английском)

Ключевые слова

social network, ensemble, bot detection, graph embedding, stacking

Издание

2nd International Workshop on Rumours and Deception in Social Media (RDSM)

Научная группа

Информационные системы

Все публикации за 2018 год Все публикации