Comparing Moderation Strategies in Group Chats with Multi-User Chatbots

Abstract

The increasing capabilities of chatbots will become more and more important in the near future. In this paper, we introduce a conversational system which connects a chatbot with a mainstream messaging service in a multi-user scenario. While there are already numerous options for single-user bots, ready-to-use systems for multiple users and group chats are scarce. This work thus aims to get insight into how such a group chatbot should behave during a multi-turn conversation. For this, we implemented and evaluated four different moderation strategies in an everyday use-case: the planning and negotiation of a joint appointment. In our subsequent user study with 40 participants, we investigated how the different strategies were perceived and what influence they had on the acceptance, usability and efficiency of the system. Our evaluation results show that users’ perceptions of innovation and inventiveness of the bot were influenced by the moderation strategies.