Creating a Bipartite Graph for a User-Item Dataset

In a content-based approach to recommendation, a lot of information is available for both items and users which is useful to create profiles. We used a graph model to represent these profiles, connecting each item to its features and each user to features of interest. Even the nearest neighbor network was built using only this information. The collaborative filtering approach, on the other hand, relies on data related to the different kinds of interactions between users and items. Such information is generally referred to as a user–item dataset.