Comparing the effectiveness of intentional preferences versus preferences over specific choices: a user study

Abstract

Many works have proposed the enrichment of database query languages with preferences, aiding the user to better rank the results. In this work we propose and examine the hypothesis that effective preference specification in many cases presupposes knowledge of the information space and the available choices. Otherwise, the expression of intentional preferences can be a tiresome process, leading to non-optimal results. We designed a user study from an information systems perspective, where participants had to express their preferences for buying a new car (intentional preferences) and then select the most preferred car from a list (preferences over specific choices). The results showed that only 20% of the users expressed intentional preferences that led to the finally selected car. The conducted statistical hypothesis testing supports the results with a 1% error. Consequently, we argue that the ability to gradually express preferences while exploring the available choices is beneficial for effectively and efficiently ranking a set of choices.