Graph-Based Entity-Oriented Search: A Unified Framework in Information Retrieval

Modern search engines have evolved beyond document retrieval. Nowadays, the information needs of the users can be directly satisfied through entity-oriented search, by taking into account the entities that better relate to the query, as opposed to relying exclusively on the best matching terms. Evolving from keyword-based to entity-oriented search poses several challenges, not only regarding the understanding of natural language queries, which are more familiar to the end-user, but also regarding the integration of unstructured documents and structured information sources such as knowledge bases. One opportunity that remains open is the research of unified frameworks for the representation and retrieval of heterogeneous information sources. The doctoral work we present here proposes graph-based models to promote the cooperation between different units of information, in order to maximize the amount of available leads that help the user satisfy an information need.