A survey on knowledge-aware news recommender systems

Abstract

News consumption has shifted over time from traditional media to online platforms, which use recommendation algorithms to help users navigate through the large incoming streams of daily news by suggesting relevant articles based on their preferences and reading behavior. In comparison to domains such as movies or e-commerce, where recommender systems have proved highly successful, the characteristics of the news domain (e.g., high frequency of articles appearing and becoming outdated, greater dynamics of user interest, less explicit relations between articles, and lack of explicit user feedback) pose additional challenges for the recommendation models. While some of these can be overcome by conventional recommendation techniques, injecting external knowledge into news recommender systems has been proposed in order to enhance recommendations by capturing information and patterns not contained in the text and metadata of articles, and hence, tackle shortcomings of traditional models. This survey provides a comprehensive review of knowledge-aware news recommender systems. We propose a taxonomy that divides the models into three categories: neural methods, non-neural entity-centric methods, and non-neural path-based methods. Moreover, the underlying recommendation algorithms, as well as their evaluations are analyzed. Lastly, open issues in the domain of knowledge-aware news recommendations are identified and potential research directions are proposed.

The RDF2vec family of knowledge graph embedding methods

“…For example, two movies which have the same director and some overlapping cast can be considered similar. This can be used, e.g., in recommender systems [22] or other predictive modeling tasks.…”

“…The standard user can directly download the gold standard and use the evaluation framework. To test class separability, the evaluation framework currently runs six machine learning classifiers which are commonly used together with embedding methods for node classification 22 (1) decision trees, (2) naïve Bayes, (3) KNN, (4) SVM, (5) random forest, and (6) a multilayer perceptron network. The framework uses the default configurations of the sklearn library.…”

NewsRecLib: A PyTorch-Lightning Library for Neural News Recommendation

“…In the past decade, numerous frameworks for the development and comprehensive evaluation of recommender systems have been proposed to address the problem of reproducibility in the field (Gantner et al, 2011;Ekstrand et al, 2011;Ekstrand, 2020;Guo et al, 2015;Kula, 2017;Da Costa et al, 2018;Salah et al, 2020;Hug, 2020;Sun et al, 2020;Anelli et al, 2021). News recommendation poses different challenges for practitioners in comparison to recommendation in domains such as movies, music, or e-commerce (Raza and Ding, 2022;Iana et al, 2022). However, few of the existing and widely used libraries offer support for news recommenders, and especially for the modern neural news recommendation models.…”

Recommendation Systems: A Systematic Review

“…In their survey of news recommenders, Iana et al (2022) stated that a "…lack of explicit user feedback …pose additional challenges for the recommendation models (p.1). This statement is a powerful observation and explains the current trend in the area of recommender systems, whereby more and more systems are valuing the incorporation of user feedback.…”