Accuracy, Exposure, and Serendipity in Movie Recommendation
Resumo
Accuracy-oriented recommenders concentrate exposure on a small fraction of the catalog. On a large Letterboxd–IMDb dataset we factor the pipeline into representation and decision policy and vary one factor at a time. With the policy held fixed at inner-product ranking, none of five representations exposes more than 14% of the catalog, and KGCN attains the best Recall@50. Holding the representation fixed at KGCN and changing only the policy takes Coverage@50 from 0.14 to 0.65, and the same ablation replicates on a second representation. Exposure is therefore governed mainly by the decision stage, and the mechanism is measurable: item embedding norms correlate with popularity (ρ up to 0.97), so inner-product ranking carries a popularity prior that cosine removes and ε-greedy bypasses. We recommend KGCN with inner-product ranking and tail exploration: it trades 9% of its recall for 4.4× the coverage, Pareto-dominates every other representation, and preserves substantially more recall than the cosine alternatives. Decomposing Seren@K into hit rate and surprise-per-hit shows that rewarding surprise raises surprise-perhit but lowers the hit rate, so measured serendipity falls. That paradox is partly an artifact of offline evaluation.
Referências
Ge, M., Delgado-Battenfeld, C., and Jannach, D. Beyond accuracy: Evaluating recommender systems by coverage and serendipity. In Proceedings of the Fourth ACM Conference on Recommender Systems. pp. 257–260, 2010.
Gilotte, A., Calauzènes, C., Nedelec, T., Abraham, A., and Dollé, S. Offline A/B testing for recommender systems. In Proceedings of the 11th ACM International Conference on Web Search and Data Mining (WSDM). pp. 198–206, 2018.
He, X., Deng, K., Wang, X., Li, Y., Zhang, Y., and Wang, M. LightGCN: Simplifying and powering graph convolution network for recommendation. In Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. pp. 639–648, 2020.
Herlocker, J. L., Konstan, J. A., Terveen, L. G., and Riedl, J. T. Evaluating collaborative filtering recommender systems. ACM Transactions on Information Systems 22 (1): 5–53, 2004.
Hu, Z., Dong, Y., Wang, K., and Sun, Y. Heterogeneous graph transformer. In Proceedings of The Web Conference 2020 (WWW). pp. 2704–2710, 2020.
Internet Movie Database. IMDb non-commercial datasets. [link], 2025.
Järvelin, K. and Kekäläinen, J. Cumulated gain-based evaluation of IR techniques. ACM Transactions on Information Systems 20 (4): 422–446, 2002.
Klimashevskaia, A., Jannach, D., Elahi, M., and Trattner, C. A survey on popularity bias in recommender systems. User Modeling and User-Adapted Interaction, 2024. arXiv:2308.01118.
Kotkov, D., Wang, S., and Veijalainen, J. A survey of serendipity in recommender systems. Knowledge-Based Systems vol. 111, pp. 180–192, 2016.
Krichene, W. and Rendle, S. On sampled metrics for item recommendation. In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. pp. 1748–1757, 2020.
Letterboxd Limited. Letterboxd — Social film discovery. [link], 2026.
Li, L., Chu, W., Langford, J., and Schapire, R. E. A contextual-bandit approach to personalized news article recommendation. In Proceedings of the 19th International Conference on World Wide Web (WWW). pp. 661–670, 2010.
Li, L., Chu, W., Langford, J., and Wang, X. Unbiased offline evaluation of contextual-bandit-based news article recommendation algorithms. In Proceedings of the 4th ACM International Conference on Web Search and Data Mining (WSDM). pp. 297–306, 2011.
Rendle, S., Freudenthaler, C., Gantner, Z., and Schmidt-Thieme, L. BPR: Bayesian personalized ranking from implicit feedback. In Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence (UAI). pp. 452–461, 2009.
Schlichtkrull, M., Kipf, T. N., Bloem, P., van den Berg, R., Titov, I., and Welling, M. Modeling relational data with graph convolutional networks. In Proceedings of the 15th Extended Semantic Web Conference (ESWC). pp. 593–607, 2018.
Shani, G. and Gunawardana, A. Evaluating recommendation systems. In Recommender Systems Handbook, F. Ricci, L. Rokach, B. Shapira, and P. B. Kantor (Eds.). Springer, Boston, MA, pp. 257–297, 2011.
Silva, T. C. and Santos, G. T. Letterboxd sample Apr 2026. [link], 2026. Kaggle dataset, CC BY-NC 4.0.
Wang, H., Zhao, M., Xie, X., Li, W., and Guo, M. Knowledge graph convolutional networks for recommender systems. In Proceedings of The World Wide Web Conference 2019 (WWW). pp. 3307–3313, 2019.
