Towards Modelling Self-imposed Filter Bubbles in Argumentative Dialogue Systems

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 4126–4134 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 4126 Towards Modelling Self-imposed Filter Bubbles in Argumentative Dialogue Systems Annalena Aicher1, Wolfgang Minker1, Stefan Ultes2 1 Institute of Communications Engineering, Ulm University, Albert-Einstein-Allee 43, 89075 Ulm 2Mercedes-Benz Research and Development, Sindelfingen, Germany annalena.aicher@uni-ulm.de Abstract To build a well-founded opinion it is natural for humans to gather and exchange new arguments. Especially when being confronted with an overwhelming amount of information, people tend to focus on only the part of the available information that fits into their current beliefs or convenient opinions. To overcome this “self-imposed filter bubble” (SFB) in the information seeking process, it is crucial to identify influential indicators for the former. Within this paper we propose and investigate indicators for the the user’s SFB, mainly their Reflective User Engagement (RUE), their Personal Relevance (PR) ranking of content-related subtopics as well as their False (FK) and True Knowledge (TK) on the topic. Therefore, we analysed the answers of 202 participants of an online conducted user study, who interacted with our argumentative dialogue system BEA (“Building Engaging Argumentation”). Moreover, also the influence of different input/output modalities (speech/speech and drop-down menu/text) on the interaction with regard to the suggested indicators was investigated. Keywords: Confirmation Bias, Spoken Dialogue Systems (SDS), Computational Argumentation, Human-Computer In- teraction (HCI), User Study, User Modelling 1. Introduction Conversations display a natural way for humans to re- solve different points of view and build an opinion. Most popular virtual agents are trained to handle sim- ple conversations, e.g. travel inquiries, still they are inept to demanding conversations (Saha et al., 2020). Especially, dialogue systems that exchange arguments and can converse with humans via natural language display a big challenge in artificial intelligence. An- other challenge displays the tendency of people which tend to focus on a biased subset of sources that re- peat or strengthen an already established or convenient opinion (Pariser, 2011). In order to avoid this pro- cess of intellectual isolation, we investigate and ana- lyze possible indicators for a self-imposed filter bubble (SFB) (Ekstr¨om, 2021). Our user study gives an in- sight into which indicators could be suitable to model an SFB and thus, provides a first step towards our aim to break the former in an engaging argumentative dia- logue. Hence, the user shall be able to scrutinize argu- ments on both sides of a controversial topic in a natu- ral and intuitive way. To this end our system engages in a cooperative dialogue with a user in order to sup- port an unbiased and critical reflected opinion building process. In particular, we focus on four main indica- tors: Reflective User Engagement (RUE), Personal Rel- evance (PR), True Knowledge (TK) and False Knowl- edge (FK). The RUE describes the critical-thinking and open-mindedness demonstrated by the user, following our definition in previous work (Aicher et al., 2021a). The PR refers to the user individual assessment of the …

Content truncated to 500 words