ISP/PhD Comprehensive Examination

. As suggested by Konstan and Riedl [66], user-contributed content is one of the vast and important data resources in recommender systems. It includes users’ tags, reviews, playlists, blog posts and micro-blogging content, forums, etc. Although the aforementioned list include the mostly agreed upon forms of user-contributed content, an extreme view can even consider ratings of users or the trust links between them as a form of user-contributed information. Although link structure can be extracted from some of the user-contributed content (either explicit such as following in Twitter or implicit such as the structure extracted from retweets), I focus on the more recognized forms of user-contributed content. To categorize user-contributed content in recommender systems, we have multiple options: based on the type of user-contributed content, the type of recommendation algorithm they are used in, the ways it is incorporated in recommender systems, and the items being recommended. Since the main focus of the question is on user-contributed contents, I choose the type of content as my main categorization factor. However, I briefly overview the other factors in the following. As for type of recommendation algorithm that user-contributed content are used in, it can be utilized in content-based [45], collaborative filtering [79], knowledge-based [16], and hybrid recommender systems [5]. Since recommender systems in general have the same categorization, I skip