Keywords Based Semantic Search

Although the semantic search has been proposed for many years, to issue semantic search query, the end users have to learn the knowledge of the underlying domain ontology and the knowledge of formal logic. On the other way, most of the web users are used to traditional keywords query even most of the search engine provide additional Boolean operator to express more complicated logics. Therefore, if we apply the keywords query to semantic search, the general web users can easily obtain the accuracy of semantic search by typing keywords.In this paper, we propose an approach which can automatically interpret users' keywords query into a ranked list of formal logic queries for semantic search. In addition, a portal type system named "SPARK" was implemented. To bridge the great gap between keywords query and formal logic based semantic search queries, we have three problem: a) Vocabulary Gap: Casual web users usually have no knowledge of the underlying ontology, so the words in their query may be quite different from those in the ontology. b) Missing of relation: keywords query does not hold relationship between teams. c) Query ranking: Due to the ambiguity of keywords query, there may be multiply formal queries produced from keyword query. How to rank these queries is a big challenge. In our spark system, the end users can type any keywords to represent his information needs, and the system can generated a ranked list of SPARQL queries for him, otherwise the user can directly submit the SPARQL query to the query engine of semantic search.The evaluation showed, the interpretation approach can achieve good performance,which also prove that natural language and keyword interface are effective approach for semantic search. In addition, we also analyze the limitations of our approach and discuss on future work.