QUICK: Expressive and Flexible Search over Knowledge Bases and Text Collections

Recent work on Web-extracted data sets has produced an interesting new source of structured Web data. These data sets can be viewed as knowledge bases (KB) – large heterogeneous linked entity collections with millions of unique edge and node labels, often encoding rich semantic information over entities. For example, YAGO [5] and ExDB [2] have fact collections numbering in the tens and hundreds of millions respectfully, and WebTables [1] contains over one hundred million extracted relations. In terms of schema information, the ExDB, YAGO, and WebTables data sets all have schema items numbering in the millions. Due to the sheer size of the schema information in these Web-extracted KBs, there has been limited development of novel tools and technologies to expose these unique data sets to end users. One of the main factors impeding the development of useful applications over these data sets is the information overload problem caused by the massive heterogeneous schema information. With KB schema items numbering in the millions, formulating structured queries is a daunting task for an application developer. Most applications are therefore built on keyword search, which does not allow direct access to the expressive structure of the data, or are applications built on structured queries with the assumption that the users are able to digest these massive schemas. As an example, consider a user searching for all scientists that have won a nobel award. Exploiting the structure of a knowledge base, a user may formalize the information need as the following SPARQL query.