YAYan Andrade

Yan Andrade

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Yan Andrade has a total of 27 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Quando Comentários e Notas Discordam: Implicações para Sistemas de Recomendação Baseados em Reviews

    by Yan Andrade, Naan Vasconcelos, Adriano Pereira, Leonardo Rocha

    Este trabalho investiga a relação entre notas e comentários no contexto de recomendação, avaliando inconsistências entre essas duas formas de interação, o potencial de notas inferidas a partir do texto e o impacto efetivo dos comentários em modelos Review-Aware Recommender Systems (RARs). Para isso, avaliamos três modelos estado-da-arte em três coleções amplamente utilizadas e empregamos a versão zero-shot do BERT para converter comentários em valores numéricos na mesma escala das notas originais. Os resultados mostram que há discrepâncias relevantes entre nota e comentário, que notas inferidas a partir do texto podem recuperar mais itens relevantes e que os RARs avaliados permanecem fortemente dependentes da nota original.


  • Open Access

    Uma Abordagem Híbrida para Sistemas de Recomendação com Base em Avaliações Textuais e Filtragem Colaborativa

    by Naan Vasconcelos, Yan Andrade, Adriano Pereira, Leonardo Rocha

    Sistemas de recomendação (SsR) utilizam cada vez mais avaliações textuais para aprimorar a modelagem de preferências. Embora os SsR baseados em revisões (RARs) capturem nuances semânticas, apresentam dificuldades em lidar com padrões colaborativos. A filtragem colaborativa (FC) tradicional é robusta em relações estruturais, mas limitada na semântica. Propomos uma abordagem híbrida que integra representações semânticas derivadas de RARs em modelos de FC. Avaliada em três conjuntos de dados da Amazon, nossa proposta demonstra que os modelos híbridos superam os modelos de referência, alcançando um erro de previsão até 2,8 vezes menor e uma precisão 60% maior, destacando seu potencial para recomendações robustas e contextualizadas.

  • STEval: A framework for evaluating spatio-temporal crime prediction models

    by Gabriel Amarante, Matheus Pimenta, Yan Andrade, Matheus Senna, Rainer Menezes, Antônio Hot Faria, Marcelo Vilas-Boas, Frederico Martins de Paula Neto, João Paulo da Silva, Everton Renato de Sousa, et al.


  • Open Access

    Review-Aware Recommender Systems (RARSs): Recent Advances, Experimental Comparative Analysis, Discussions, and New Directions

    by Guilherme Bittencourt, Naan Vasconcelos, Yan Andrade, Nícollas Silva, Washington Cunha, Diego Roberto Colombo Dias, Marcos André Gonçalves, Leonardo Rocha

    Recommender systems (RSs) have emerged as an effective strategy for dealing with information overload. Their importance is undeniable as they are widely adopted in various web applications, presenting the potential to solve various problems associated with the abundance of choices. In recent years, the literature has witnessed the proposal of numerous recommendation techniques, constantly seeking to improve the effectiveness of these methods. Although many approaches have focused on techniques that employ quantitative (numerical) evaluations, it is essential to recognize the significant value of the user feedback , usually expressed as (textual) comments (a.k.a., reviews) in order to understand their preferences. In this context, various algorithms have been developed to effectively take advantage of comments as a valuable source of information for Review-Aware RSs (RARSs), capable of generating recommendations in line with each user’s profile using their respective comments. This article aims to review the primary research efforts on RARSs comprehensively. We present a taxonomy of recommendation models in this domain, accompanied by a detailed summary of the main advances, the main datasets, and the used metrics. In addition, we conduct a comprehensive experimental comparison among state-of-the-art proposals, highlighting the main directions and new perspectives for future developments. Code and datasets of our open benchmark are available at https://github.com/guibitten03/iRev for fostering replicability and new advances.


  • Open Access

    Um Framework para Avaliação de Modelos de Previsão Espaço-Temporal de Crimes

    by Yan Andrade, George Teodoro, Renato Ferreira, Leonardo Rocha

    Modelos preditivos espaço-temporais são fundamentais para prever locais e momentos de crimes, auxiliando a segurança pública. No entanto, a ausência de critérios padronizados de avaliação dificulta a comparação entre diferentes abordagens. Para superar essa limitação, apresentamos o STEval, um framework flexível de avaliação. A robustez do STEval foi validada por meio de experimentos que variaram a granularidade temporal e espacial. Os resultados indicaram que nenhum dos modelos supera os demais em todos os cenários. O framework auxilia na seleção e otimização de modelos por meio de análises detalhadas. Além disso, o STEval expõe as forças e fraquezas de cada modelo, incentivando avanços na pesquisa sobre previsão de crimes.

  • Open Access

    A Survey on Review - Aware Recommendation Systems

    by Guilherme Bittencourt, Guilherme Fonseca, Yan Andrade, Nícollas Silva, Leonardo Rocha

    Current advances in Deep Neural Networks have motivated recent studies to return their interest in Review-Aware Recommender Systems (RSs). In this sense, we employ a systematic mapping approach by selecting 56 papers published on the main vehicles of the area, such as RecSys, VLDB, SIGIR, and others. Reading these works and synthesising their achievements, we provide an updated picture of this field by highlighting relevant outcomes, contributions, and limitations. Especially, we have identified two huge limitations in the area. First, many of these works do not provide their code sources. Second, there is a prevalent tendency to prioritize accuracy above other quality dimensions, despite the consensus within the community to assess the practical effectiveness of RSs through many metrics. Both observations create constraints that restrict reproducibility and hinder straightforward comparisons in the field. Addressing this gap, this work also provides an open-source library that compiles the principal strategies proposed in the literature. Furthermore, we conducted a comprehensive evaluation through several datasets and metrics. Our findings indicate that while no single algorithm demonstrates absolute superiority, RSs based on neural networks have particularly exhibited the most competitive performance.


  • Open Access

    A Complete Framework for Offline and Counterfactual Evaluations of Interactive Recommendation Systems

    by Yan Andrade, Nícollas Silva, Thiago Silva, Adriano Pereira, Diego Dias, Elisa T. Albergaria, Leonardo Rocha

    Interactive recommendation has been recognized as a Multi-Armed Bandit (MAB) problem. Items are arms to be pulled (i.e., recommended) and the user’s satisfaction is the reward to be maximized. Despite the advances, there is still a lack of consensus on the best practices to evaluate such solutions. Recently, two complementary frameworks were proposed to evaluate bandit solutions more accurately: iRec and OBP. The first one has a complete set of offline metrics and bandit models that allows us to perform an comparisons with several evaluation policies. The second one provides a huge set of bandit models to be evaluated through several counterfactual estimators. However, there is a room to be explored when joining these two frameworks. We propose and evaluate an integration between both, demonstrating the potential and richness of such combination.