Automating the Discovery of Recommendation Rules

Preferences in artificial intelligence

“…This is similar to what happens in case-based reasoning (a subfield of artificial intelligence) where solutions to previous similar problems are adapted in order to solve the problem at hand. This idea has been also considered in preference-based systems, in particular in approaches to recommendation based on case-based reasoning [186,244].…”

Knowledge Discovery from User Preferences in Conversational Recommendation

“…First, the recommender computes the compatibility score, as detailed in Equation 3. It is important to note that the compatibility function has also been modified, as explained below in Equation 3. Instead of averaging the compatibility and the similarity, as is done with incremental critiquing, our second step assembles the cases with the highest compatibility from the list of remaining cases.…”

“…As part of their cyclic recommendation process, conversational systems aim to retrieve products that respect user preferences by requiring users to provide minimal feedback in each cycle. It is expected that over the course of a recommendation session that the recommender learns more about user preferences and therefore it can assist in the discovery of recommendation knowledge which prioritises products that best satisfy these preferences [2,3]. Advantages of the approach include: (1) users have more control over the navigation process [4]; and (2) users are guided to target products faster than standard browsing and alternative recommendation approaches [5,6].…”

“…As recommender systems become more commonplace, there has been renewed interest in critiquing, with the major focus of research in increasing the efficiency of recommendation dialogues [9]. Furthermore, recent research has highlighted the importance of investigating techniques for automating the discovery of implicit preference knowledge while requiring minimal information input from the user [3].…”