Searching Linked Objects with Falcons

The rapid development of the data Web is accompanied by increasing information needs from ordinary Web users for searching objects and their relations. To meet the challenge, this chapter presents Falcons Object Search, a keyword-based search engine for linked objects. To support various user needs expressed via keyword queries, for each object an extensive virtual document is indexed, which consists of not only associated literals but also the textual descriptions of associated links and linked objects. The resulting objects are ranked according to a combination of their relevance to the query and their popularity. For each resulting object, a query-relevant structured snippet is provided to show the associated literals and linked objects matched with the query for reflecting query relevance and even directly answering the question behind the query. To exploit ontological semantics for more precise search results, the type information of objects is leveraged to support class-based query refinement, and Web-scale class-inclusion reasoning is performed to discover implicit type information. Further, a subclass recommendation technique is proposed to allow users navigate class hierarchies for incremental results filtering. A task-based experiment demonstrates the promising features of the system.