Providing personalized recommendation for attending events based on individual interest profiles
Boris A. Galitsky · Artificial Intelligence Research · 2015
In this article we present a framework to extract user interests from social network profiles such as Facebook to personalize recommendations about products and services. Matching users’ interests as keywords with product attributes as keywords, performed by currently available personalization systems, has a very low recall, so more general category-based frameworkis needed. It turns out that substantial reasoning about products and their categories is required to match a taxonomy of theowner of products and services, with that of a user, as expressed in a public profile. To handle inconsistencies between the setaxonomies, a mapping of one into another is expressed as a Defeasible Logic program (DeLP), where a potential mapping can bedefeated by other ones if relevant information becomes available. Events and things to do are recommended at StubHub.com and www.facebook.com/StubHub/ so that the reader can observe the system at a scale. Also, we present content management system which supports personalized recommendation is outlined.