ScholarWorks@성균관대학교

초록

Toxic and aggressive comments are a persistent problem in online spaces, and platforms commonly rely on blocking-based moderation to address them. However, blocking is only one way of intervening during comment writing, and alternative approaches that aim to preserve user expression have rarely been examined in direct comparison. This study aims to compare blocking and refinement-based moderation during comment writing to assess how they differentially shape user behavior, user perceptions, and semantic properties of the resulting comments. We deployed a web-based commenting platform and conducted a controlled user experiment in which moderation intervened in real time. Our results show that refinement primarily affects the semantic and emotional characteristics of comments, whereas blocking influences users' evaluative responses to the moderation experience. These findings clarify that moderation strategies intervene in different dimensions of the commenting process and provide a basis for designing moderation systems that balance textual outcomes and user experience.

키워드