Loading...

Probabilistic Graphical models in Recommender systems

Seyed Aboutorabi, Hediyeh Sadat | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57406 (02)
  4. University: Sharif University of Technology
  5. Department: Mathematical Sciences
  6. Advisor(s): Moghadasi, Reza
  7. Abstract:
  8. In recommender systems, the available data structure is mostly in the form of a graph; Therefore, during the past years, graph based recommendation methods, especially Graph neural networks, have received a lot of attention. These networks obtain the representation of user nodes and items by aggregating multi-hop neighbors in different ways. Unfortunately, the existing methods are part of the Supervised learning scenario and due to the lack of data in recommender systems, they have faced serious challenges. These challenges reduce the ability to learn to representation nodes and the accuracy of recommendations. Recently, knowledge graph based recommendations have achieved excellent performance with the aim of incorporating knowledge graph as auxiliary information. Especially if the interaction between users and items is modeled by means of graphic networks. In this thesis, we discuss two limitations of these models, one is the unbalanced use of the information in the graph and the other is the dispersion of the supervised signal. After that, by applying self-supervised learning techniques and generating three views of the knowledge graph combined with the user-item interaction graph, we design the proposed model to solve these two challenges. The experimental results on the three evaluation criteria Recall, AUC and F1 show the superiority of the proposed model compared to other advanced methods. Also, this model is among the first models that used the multi-level contrastive learning mechanism in knowledge graph-based recommendations
  9. Keywords:
  10. Recommender System ; Graphical Methods ; Knowledge Graph ; Self-Supervised Learning ; Graph Neural Network

 Digital Object List

 Bookmark

...see more