Structured Querying of Annotation-Rich Web Text with Shallow Semantics

Information discovery on the Web has so far been dominated by keyword-based document search. However, recent years have witnessed arising needs from Web users to search for named entities, e.g., finding all Silicon Valley companies. With existing Web search engines, users have to digest returned Web pages by themselves to find the answers. Entity search has been introduced as a solution to this problem. However, existing entity search systems are limited in their capability to address complex information needs that involve multiple entities and their interrelationships. In this report, we introduce a novel entity-centric structured querying mechanism called Shallow Semantic Query (SSQ) to overcome this limitation. We cover two key technical issues with regard to SSQ, ranking and query processing. Comprehensive experiments show that (1) our ranking model beats state-of-the-art entity ranking methods; (2) the proposed query processing algorithm based on our new Entity-Centric Index is more efficient than a baseline extended from existing entity search systems.