Efficient Exploration of Linked Data

Harnessing the potential of the Semantic Web for building knowledgeable machines entails the ability to understand RDF graphs and integrate them with applications. Analyzing the vast information using common tools requires skill and time. Towards that we develop ELinda - an explorer for linked data. ELinda enables the understanding of the rich content stored in an RDF graph, via a visual query language for interactive exploration. The focus is on rich and open-domain datasets where it is especially challenging to detect the precise value to the application at hand. In essence, the model is based on the concept of a bar chart that depicts the distribution of a focus set of nodes (URIs), and each bar can be expanded to a new bar chart for further exploration. Three types of expansions are supported: subclass, property, and object. Under the hood, our visual query language is compiled into SPARQL. Yet, these queries require prohibitively long execution times on a standard SPARQL engine. To address this challenge, we develop a specialized query engine that is based on the concept of a worst-case-optimal join algorithm. The novel query engine provides a speedup of 1-2 orders of magnitude compared to standard SPARQL engines, and thereby facilitates the practical implementation of ELinda.