Representation of rules for relevant recommendations to online social networks users

Abstract

In our prior work, we identified rules for use in recommendation algorithms on Online Social Network (OSN) in order to increase the relevance of content suggested to a user. The resulting recommendation algorithms filter out and prioritize event types for OSN users (such as photo posts by friends, status posts, shared content, etc.), and are thereby intended to reduce information overload. This paper proposes a representation of these rules in a requirements model of a OSN. This is interesting, because recommendation rules influence user behavior, which in turn influences future requirements. If there is a recommendation algorithm, then its behavior should be represented also in requirements models of the system. The paper makes two contributions. We define requirements that OSNs should satisfy in order to produce relevant recommendations of event types to users. We investigate whether an existing requirements modeling language (namely, i-star) can be used to model these requirements.