News Aggregator with Recommendation System Based on Reliability of Sources

Reading news online has become extremely popular as the Internet gives access to news articles on various topics from thousands of different sources around the world. To make this process more convenient there are so-called news aggregation services. A key challenge of any such service is to help users to find news articles that are interesting to read and to save them time by concealing articles that probably they will not like. However the modern world faced a challenge when the spread of deceptive information in common media sources such as online newspapers, social networks news feeds, and blogs have made it difficult to determine reliable news sources, thus increasing the requirement for tools able to assess the trustworthiness of online content. This work is focused on developing a news aggregation service with a personalized recommendation system. The recommendation system was implemented using a developed Bayesian model to predict users’ news preferences based on their likes and dislikes data. It ranks recommended news according to the reliability index of the source, which is calculated using natural language processing and cluster analysis tools. The platform is built on top of modern web stack technologies including Node.js, Python, GraphQL, Prisma, and React. It allows users to explore new news articles and news publishers, helping users finding sources that will be suitable for their interests from trustworthy sources.