Entity-centric search: querying by entities and for entities

The immense scale of the Web has rendered itself as a huge repository storing information about various types of entities (e.g., persons, locations, products, companies). Much of information retrieval operations on the Web nowadays are about entities, i.e., entity-centric, for example, finding cameras that have black color and high resolution, keeping track of important events of a favorite celebrity, etc.. However, without modeling the concept of entities, most search engines still take the page view of the Web data, which accepts keyword queries as input, and aims at finding documents that are relevant to the queries. Therefore, in my thesis, I propose to study entity-centric search towards facilitating various types of entity-related information search operations. When the concept of entity is involved in a search operation, people are usually interested in finding more information about some known entities, or exploring unknown entities that satisfy certain information needs. I propose to categorize such operations into two categories: querying by entities and querying for entities. As the objective of my thesis, I aim at building a general framework to facilitate these different types of entity-centric search operations. First, in query by entities, I propose to address the entity-centric document filtering problem, which, towards better characterizing the target entity, adopts its entity identification page (e.g., Wikipedia page) as input to identify its relevant documents. Second, in query for entities, witnessing many different ad-hoc efforts for exploiting fine granularity entity information across Web text, e.g., typed-entity search, question answering, I propose to build a general data-oriented content query system, which distills their essential capabilities and supports ``content querying'' for finding various entity data in the text. Third, I study the relational entity search problem, i.e., given a query entity, how to search entities that match a desired relation -- since the search operation involves entities in both input query and output result, it belongs to querying by entities and for entities at the same time. The results we obtained so far show clear promise of entity-centric search in its usefulness, effectiveness and efficiency