Feds: A Highly Efficient Keyword Search System Operating on Federated RDF Systems

In this demonstration, we develop Feds, a federated RDF-based keyword search system. Our approach involves constructing an offline schema graph which represents the structure of the federated RDF system. By mapping user-inputted keywords to the vertices in the schema graph, we establish a connection between the keywords and the underlying RDF data. Once the keyword-to-vertex mapping is established, we leverage the schema graph to transform the user’s keywords into SPARQL queries. The constructed queries are then executed over the federated RDF system, allowing us to obtain the desired search results. To measure the efficiency and effectiveness of our system, we propose cost models that evaluate the performance of keyword mapping and query construction. To further optimize the evaluation process, we introduce a multiple query optimization strategy. Through extensive experimental studies using real-world federated RDF datasets, our demonstration shows remarkable efficiency, effectiveness, and scalability in performing keyword search over federated RDF systems.