More and more RDF data have been published online to be consumed. Ordinary Web users also expect to experience more intelligent services promised by the Semantic Web, such as object search based on structured data. We implemented the Falcons search engine to meet the challenge. To enable keyword search, for each object, we construct and index a virtual document that includes textual descriptions of its neighboring resources. Typing information is used to serve class-based query refinement, and class-inclusion reasoning is performed to discover implicit types of objects. A method of recommending subclasses is implemented to enable navigating class hierarchies for incremental query refinement. We also report on lessons learned from Web-scale experiments.
展开▼
