Keyword Search over RDF Graphs Using WordNet.

An increasing amount of interlinked RDF datasets are published on the Web. These datasets can be queried using languages such as Sparql, in which graph patterns are evaluated against the data. In such languages, some knowledge about the dataset is required in order to formulate a query, such as the resources, types or properties existing in the dataset. An alternative way of querying RDF data is keyword search, which could be very useful when the content of the dataset is not known. One of the problems we are faced with is the gap between the keywords of the query and the terms used in the dataset. In this paper, we introduce a keyword search approach for RDF data which aims at solving this terminological gap. Our approach makes use of external knowledge provided by online linguistic resources and proposes a ranking method if several results are returned for a given query. We have performed some experiments on the DBpedia and the AIFB datasets to illustrate the scalability and efficiency of our approach.