Authors: (1) Yi-Ling Chung, The Alan Turing Institute (ychung@turing.ac.uk); (2) Gavin Abercrombie, The Interaction Lab, Heriot-Watt University (g.abercrombie@hw.ac.uk); (3) Florence Enock, The Alan Turing Institute (fenock@turing.ac.uk); (4) Jonathan Bright, The Alan Turing Institute (jbright@turing.ac.uk); (5) Verena Rieser, The Interaction Lab, Heriot-Watt University and now at Google DeepMind (v.t.rieser@hw.ac.uk). Authors: Authors: (1) Yi-Ling Chung, The Alan Turing Institute (ychung@turing.ac.uk); (2) Gavin Abercrombie, The Interaction Lab, Heriot-Watt University (g.abercrombie@hw.ac.uk); (3) Florence Enock, The Alan Turing Institute (fenock@turing.ac.uk); (4) Jonathan Bright, The Alan Turing Institute (jbright@turing.ac.uk); (5) Verena Rieser, The Interaction Lab, Heriot-Watt University and now at Google DeepMind (v.t.rieser@hw.ac.uk). Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 Background 2 Background 3 Review Methodology 3 Review Methodology 4 Defining counterspeech 4 Defining counterspeech 4.1 Classifying counterspeech 4.1 Classifying counterspeech 5 The Impact of Counterspeech 5 The Impact of Counterspeech 6 Computational Approaches to Counterspeech and 6.1 Counterspeech Datasets 6 Computational Approaches to Counterspeech and 6.1 Counterspeech Datasets 6.2 Approaches to Counterspeech Detection and 6.3 Approaches to Counterspeech Generation 6.2 Approaches to Counterspeech Detection and 6.3 Approaches to Counterspeech Generation 7 Future Perspectives 7 Future Perspectives 8 Conclusion, Acknowledgements, and References 8 Conclusion, Acknowledgements, and References 3 Review Methodology Taking a multi-disciplinary perspective, we systematically review work on counterspeech from computer science and the social sciences published in the past ten years. To ensure broad coverage and to conduct a reproducible review, we follow the systematic methodology of Moher et al. (2009). The search and inclusion process is shown in Figure 2. We used keyword terms related to counterspeech to search three key databases (ACL Anthology, ArXiv, and Scopus) that together offer a broad coverage of our target literature. We included the search terms ‘counter-speech’, ‘counter-narratives’, ‘counter-terrorism’, ‘counter-aggression’, ‘counter-hate’, ‘counter speech’, ‘counter narrative’, ‘countering online hate speech’, ‘counter hate speech’, and ‘counter-hate speech’. We also included 34 publications that we had identified previously from other sources, but that were not returned by keyword search due to not including relevant keywords or not being indexed in the target search repositories. The search was conducted in December 2022. Of the returned results, we include all publications that concern (1) analysis of the use and effectiveness of interventions against hateful or abusive language online, (2) characteristics of counterspeech users or recipients, or (3) data and/or implementation designed for counterspeech (e.g., counterspeech classification or generation). These inclusion criteria were applied by two of the authors. Following this process, we include 90 papers for analysis in this review. Each of the papers was read by at least one of the co-authors of the article. Our review is divided into several sections (the results of which are presented sequentially below). First, we examine definitional characteristics of counterspeech, looking at how the term itself is defined, how different taxonomies have been created to classify different types of counterspeech, and the different potential purposes attributed to it. Then, we examine studies that have looked at the impact of counterspeech, discussing the different analytical designs employed and analysing evidence of the results. Following this, we discuss computational approaches to counterspeech, focusing in particular on both detection and generation. Finally, we examine ethical issues in the domain of counterspeech, and also speculate about future perspectives and directions in the field. This paper is available on arxiv under CC BY-SA 4.0 DEED license. This paper is available on arxiv under CC BY-SA 4.0 DEED license. available on arxiv
