Evaluating Bi-Relational Data Representation for Collaborative Filtering
[ 1 ] Instytut Automatyki i Inżynierii Informatycznej, Wydział Elektryczny, Politechnika Poznańska | [ P ] pracownik
2014
artykuł naukowy
angielski
- Collaborative Filtering
- RDF data representation
EN Many collaborative filtering methods involve the use of an input matrix representing each user as a vector in a space of items and, analogically, each item as a vector in a space of users. In this paper we propose to use an element-fact matrix – in which columns represent RDF-like triples and rows represent users, items, and relations – in order to represent the behavioral input data of the form of (userX, likes, itemY) and (userX, dislikes, itemY) triples. Moreover, we provide the evaluation framework based on the use of AUROC measure, and two experimental settings – for the one-class and for the bi-relational collaborative filtering scenarios. One of the key findings of the research presented in this paper is that the proposed data representation scheme, while combined with reflective matrix processing, significantly outperforms state-of-the-art collaborative filtering methods based on the use of a ‘classical’ user-item matrix as the input data representation.
104 - 116
10