Abstract

In this paper we present an analysis of the social movie recommendation community Filmtipset. Filmtipset is Sweden's largest movie recommendation community with more than 80,000 users. The website offers movie recommendations based on usage data, but also consists of a social network where users are able to befriend one another. All content is user-generated and there is a multitude of features that make the dataset stand out among other movie recommendation datasets. We evaluate the social graphs' impact on users similarities in taste in movies, and show that utilizing this relation could be used to improve movie recommendation quality.

@INPROCEEDINGS{said10a,
  author = {Alan Said and Ernesto W. {De Luca} and Sahin Albayrak},
  title = {How social relationships affect user similarities},
  booktitle = {Proceedings of the ACM IUI'10 Workshop on Social Recommender Systems},
  year = {2010},
  location = {Hong Kong, China},
  owner = {said}
}