Estratégias para Aprimorar a Diversidade Categórica e Geográfica de Sistemas de Recomendação de POIs
Resumo
Currently, so-called Recommendation Systems (SRs) have been used to assist users in discovering relevant Points of Interest (POIs) on Location-Based Social Networks (LBSN), such as FourSquare and Yelp. Given the main challenges of data-sparse and geographic influence in this scenario, most of the work on POI recommendations has focused only on improving the effectiveness (i.e. accuracy) of the systems. However, there is a growing consensus that just effectiveness is not sufficient to assess the practical utility of these systems. In real scenarios, categorical and geographic diversities were identified as the main complementary dimensions for assessing user satisfaction and the usefulness of recommendations. The works in the literature are concentrated on only one of these concepts. In this work, we propose a new post-processing strategy, which combines these concepts in order to improve the user’s interest in POIs. Our experimental results in the Yelp data sets show that our strategy can improve user satisfaction, considering different SRs and multiple diversification metrics. Our method is capable of improving diversity by up to 120 % without significant losses in terms of effectiveness.
Referências
Rodrigo Carvalho, Nícollas Silva, Luiz Chaves, Adriano C. M. Pereira, and Leonardo C. da Rocha. 2019. Geographic-categorical diversification in POI recommendations. In WebMedia.
Luiz Chaves, Nícollas Silva, Rodrigo Carvalho, Adriano C. M. Pereira, and Leo-nardo C. da Rocha. 2019. Exploiting the user activity-level to improve the models’ accuracy in point-of-interest recommender systems. In WebMedia.
Van Dang and W Bruce Croft. 2012. Diversity by proportionality: an election-based approach to search result diversification. In Proceedings of the international ACM SIGIR conference on Research and development in information retrieval. ACM.
Jean-Benoît Griesner, Talel Abdessalem, and Hubert Naacke. 2015. POI recommendation: towards fused matrix factorization with geographical and temporal influences. In Proceedings of the 9th ACM RecSys. ACM, 301–304.
Jungkyu Han and Hayato Yamana. 2017. Geographical Diversification in POI Recommendation. In Proceedings of the Eleventh ACM Conference on Recommender Systems - RecSys ’17. https://doi.org/10.1145/3109859.3109884
Alessio Ishizaka and Philippe Nemery. 2013. Multi-criteria Decision Analysis: Methods and Software. Wiley. https://www.amazon.com/Multi-criteria-Decision-Analysis-Methods-Software/dp/1119974070?SubscriptionId=AKIAIOBINVZYXZQZ2U3A&tag=chimbori05-20&linkCode=xm2&camp=2025&creative=165953&creativeASIN=1119974070
Silvia Beatriz Neiva and Luiz Flavio Autran Monteiro Gomes. 2009. A aplicação da teoria da utilidade multiatributo à escolha de um software de e-procurement. Revista Tecnologia 28, 2 (2009).
Shameem A Puthiya Parambath, Nicolas Usunier, and Yves Grandvalet. 2016. A coverage-based approach to recommendation diversity on similarity graph. InRecSys. 15–22.
Saúl Vargas, Linas Baltrunas, Alexandros Karatzoglou, and Pablo Castells. 2014. Coverage, redundancy and size-awareness in genre diversity for recommender systems. In RecSys.
Saúl Vargas and Pablo Castells. 2011. Rank and relevance in novelty and diversity metrics for recommender systems. In RecSys. https://doi.org/10.1145/2043932.204395517
