Network Science @ Recommender Systems

We present a conceptual approach in the field of recommender systems, which is intended to model human consumption by maintaining a network of heterogeneous nodes and relationships. We think of this model as the reflection of the corresponding cognitive functionality of human thinking, as we maintain a structure which is similar to the structures established by neural networks. To explain our motivation and the proposed structure we are combining the results of recommender systems and network science. We propose a generalized approach that intends to involve concepts from social networks, semantic distance, association rule mining, ontological modeling and expert systems. Our approach will access and integrate different information sources, modeling also additional information types. We expect that our approach will find the importance factors of the aforementioned information sources for the generation of high quality recommendations.