Retrieving Under Uncertainty: Towards a Chatbot Uncertainty Taxonomy (CUT) for Information Retrieval
Gespeichert in:
| Titel: | Retrieving Under Uncertainty: Towards a Chatbot Uncertainty Taxonomy (CUT) for Information Retrieval |
|---|---|
| Autoren: | Alina Asisof |
| Quelle: | Proceedings of the 2025 International ACM SIGIR Conference on Innovative Concepts and Theories in Information Retrieval (ICTIR). :155-166 |
| Verlagsinformationen: | ACM, 2025. |
| Publikationsjahr: | 2025 |
| Schlagwörter: | taxonomy, framework, uncertainty, Chatbot, information |
| Beschreibung: | Using chatbots for information retrieval is characterized by uncertainty. While traditional approaches focus on system-related uncertainty, this paper emphasizes the importance of user-centric perceived uncertainties. The introduced Chatbot Uncertainty Taxonomy (CUT) outlines that users experience diverse, context-dependent uncertainties, consisting of functional, ethical, privacy-related, relational, and operational dimensions. The taxonomy is empirically explored and illustrated using a survey with 50 participants who indicate the importance of the uncertainties across four scenarios (hedonic vs. utilitarian scenario with a new/known chatbot). The results indicate that functional uncertainties prevail in all scenarios, but operational uncertainties dominate utilitarian settings, while ethical and relational uncertainties are more important in hedonic ones. Additionally, first-time interactions with yet unknown chatbots are ridden with functional and operational uncertainty, whereas for known chatbot interactions, privacy, ethical, and relational uncertainties dominate, and operational uncertainty is reduced. Concludingly, context and prior experience significantly shape user uncertainty priorities. The taxonomy contributes to the broader discussion by shifting the focus from model uncertainty to user perception and relevance, proposing a nuanced, user-focused framework for understanding and designing uncertainty-aware chatbot systems. |
| Publikationsart: | Article Conference object |
| DOI: | 10.1145/3731120.3744580 |
| DOI: | 10.3929/ethz-c-000784169 |
| Rights: | CC BY |
| Dokumentencode: | edsair.doi.dedup.....8b8dbd48fc8053beefbe6c36ee3fc68c |
| Datenbank: | OpenAIRE |
Nájsť tento článok vo Web of Science | Abstract: | Using chatbots for information retrieval is characterized by uncertainty. While traditional approaches focus on system-related uncertainty, this paper emphasizes the importance of user-centric perceived uncertainties. The introduced Chatbot Uncertainty Taxonomy (CUT) outlines that users experience diverse, context-dependent uncertainties, consisting of functional, ethical, privacy-related, relational, and operational dimensions. The taxonomy is empirically explored and illustrated using a survey with 50 participants who indicate the importance of the uncertainties across four scenarios (hedonic vs. utilitarian scenario with a new/known chatbot). The results indicate that functional uncertainties prevail in all scenarios, but operational uncertainties dominate utilitarian settings, while ethical and relational uncertainties are more important in hedonic ones. Additionally, first-time interactions with yet unknown chatbots are ridden with functional and operational uncertainty, whereas for known chatbot interactions, privacy, ethical, and relational uncertainties dominate, and operational uncertainty is reduced. Concludingly, context and prior experience significantly shape user uncertainty priorities. The taxonomy contributes to the broader discussion by shifting the focus from model uncertainty to user perception and relevance, proposing a nuanced, user-focused framework for understanding and designing uncertainty-aware chatbot systems. |
|---|---|
| DOI: | 10.1145/3731120.3744580 |