Evaluating the Unification of Multiple Information Retrieval Techniques into a News Indexing Service

While online information sources are rapidly increasing in amount, so does the daily available online news content. Several approaches have being proposed for organizing this immense amount of data. In this work we explore the integration of multiple information retrieval techniques, like text preprocessing, n-grams expansion, summarization, categorization and item/user clustering into a single mechanism designed to consolidate and index news articles from major news portals from around the web. Our goal is to allow users to seamlessly and quickly get the news of the day that are of appeal to them via our system. We show how, the application of each one of the proposed techniques gradually improves the precision results in terms of the suggested news articles for a number of registered system users and how, aggregately, these techniques provide a unified solution to the recommendation problem.