Web search calls for more sophisticated capabilities than document retrieval. A promising approach is to respond to a user query with specific objects or entities of interest. For example, a query of “dentist” will return a list of dental clinics in the area with their websites, phone numbers, hours of operation, names of doctors, locations on the map, and so on, gathered and constructed from structured and unstructured web data. This constitutes an entity search.

The paper proposes methods combining entity relevance, type estimation, type matching, entity prior, and entity co-occurrence in a unified probabilistic model. The authors conducted testing of both individual entity types and their combinations to understand contributions to quality output of each type. The results of the experiments show that the proposed method can improve search performance.

The intended audience of the paper is academics in information retrieval and the semantic web. Readers who are interested in this topic can find additional information on the subject [1,2].