Forgetting on Evolving Graphs for Accurate and Diverse Stream-Based Recommendation

• Murilo F. L. Schmitt Universidade Federal do Paraná
• Eduardo J. Spinosa Universidade Federal do Paraná

DOI: https://doi.org/10.5753/kdmile.2022.227804

Resumo

Stream-based recommender systems are an active research field, relying on incremental algorithms to update models by incorporating new data on a single pass, discarding such data after processing. A limitation of solely including new data is the accumulation of obsolete concepts, which eventually raises accuracy and scalability concerns. In this work, we propose a gradual forgetting technique for incremental neighborhood-based methods that locally forgets items based on recency and popularity, by decreasing importance of neighborhood of items for every incoming observation to emphasize more recent and reinforced ones. The technique includes parameters to increase diversity, by retaining less popular yet relevant items, and scalability, by pruning obsolete connections not reinforced by new data. Experiments conducted by extending a recent incremental graph-based approach highlight the effectiveness of the proposed technique, as its application improved scalability and diversity, outperforming baselines.

Palavras-chave: recommender systems, data streams, forgetting, online learning

Referências

Al-Ghossein, M., Abdessalem, T., and Barre, A. A survey on stream-based recommender systems. ACM Computing Surveys (CSUR) 54 (5): 1–36, 2021.

Cremonesi, P., Koren, Y., and Turrin, R. Performance of recommender algorithms on top-n recommendation tasks. In Proceedings of the 4th ACM conference on Recommender systems. pp. 39–46, 2010.

Ding, Y. and Li, X. Time weight collaborative filtering. In Proceedings of the 14th ACM international conference on Information and knowledge management, 2005.

Frigó, E., Pálovics, R., Kelen, D., Kocsis, L., and Benczúr, A. Online ranking prediction in non-stationary environments. In Proceedings of the 1st Workshop on Temporal Reasoning in Recommender Systems. ACM, 2017.

Gama, J., Žliobaite, I. ̇ , Bifet, A., Pechenizkiy, M., and Bouchachia, A. A survey on concept drift adaptation. ACM computing surveys (CSUR) 46 (4): 1–37, 2014.

Koren, Y. Collaborative filtering with temporal dynamics. In Proceedings of the 15th ACM SIGKDD, 2009.

Matuszyk, P., Vinagre, J., Spiliopoulou, M., Jorge, A. M., and Gama, J. Forgetting techniques for stream-based matrix factorization in recommender systems. Knowledge and Information Systems 55 (2): 275–304, 2018.

Ning, X., Desrosiers, C., and Karypis, G. A comprehensive survey of neighborhood-based recommendation methods. Recommender systems handbook, 2015.

Schmitt, M. F. and Spinosa, E. J. Scalable stream-based recommendations with random walks on incremental graph of sequential interactions with implicit feedback. User Modeling and User-Adapted Interaction 32 (4): 543–573, 2022.

Schmitt, M. F. L. and Spinosa, E. J. Incremental graph of sequential interactions for online recommendation with implicit feedback. In 3rd Workshop on Online Recommender Systems and User Modeling, 2020.

Smyth, B. and McClave, P. Similarity vs. diversity. In ICCBR. Springer, pp. 347–361, 2001.

Tabassum, S., Veloso, B., and Gama, J. On fast and scalable recurring link’s prediction in evolving multi-graph streams. Network Science 8 (S1): S65–S81, 2020.

Vinagre, J. and Jorge, A. M. Forgetting mechanisms for scalable collaborative filtering. Journal of the Brazilian Computer Society 18 (4): 271–282, 2012.

Vinagre, J., Jorge, A. M., and Gama, J. Collaborative filtering with recency-based negative feedback. In Proceedings of the 30th Annual ACM SAC. Spain, pp. 963–965, 2015.

Vinagre, J., Jorge, A. M., Rocha, C., and Gama, J. Statistically robust evaluation of stream-based recommender systems. IEEE Transactions on Knowledge and Data Engineering 33 (7): 2971–2982, 2019.