Factors Influencing User Motivation for Giving Online Preference Feedback

Preference elicitation is important for any computerized system advising users about choices. Recommender systems aim to propose interesting material to users. Therefore, they must first gather user preferences. Negotiation support systems can only give meaningful bidding advice based on users’ preferences regarding negotiable issues and interests. In general, the more detail users are willing to give the better the support will be. We investigated how four factors indluence users’ willingness to give detail in an online preference elicitation experiment. 18 users rated 60 items (pictures/songs) with 5 levels of detail (from 3-point scale over affective feedback to free text). For each item, users could choose the desired detail level. Our results show that, of the four factors investigated (having an opinion about an item, content type of the item, familiarity with, and ownership of that item), mainly having an opinion about an item makes users give significantly more detailed feedback. Further, users with an opinion about an item use qualitatively rich affective feedback in 30% of the cases. Our findings indicate that adaptive preference elicitation interfaces can conditionally hide and show fine grained feedback providing a simpler interface, whch can be important for smaller interfaces. Further, the fact that 30% of the cases rated with an opinion included affective detail indicates that users are willing to give rich affective feedback when they have an opinion.