LODVader

Ciro Baron Neto, Kay Müller, Martin Brümmer, Dimitris Kontokostas, Sebastian Hellmann
2016 Proceedings of the 25th International Conference Companion on World Wide Web - WWW '16 Companion  
The Linked Open Data (LOD) cloud is in danger of becoming a black box. Simple questions such as "What kind of datasets are in the LOD cloud?", "In what way(s) are these datasets connected?" -albeit frequently asked -are at the moment still difficult to answer due to the lack of proper tooling support. The infrequent update of the static LOD cloud diagram adds to the current dilemma, since there is neither reliable nor timely-updated information to perform an interactive search, analysis or in
more » ... h, analysis or in particular visualization in order to gain insight into the current state of Linked Open Data. In this paper, we propose a new hybrid system which combines LOD Visualisation, Analytics and Discov-ERy (LODVader) to aid in answering the above questions. LODVader is equipped with (1) a multi-layer LOD cloud visualization component comprising datasets, subsets and vocabularies, (2) dataset analysis components that extend the state of the art with new similarity measures and efficient link extracting techniques and (3) a fast search index that is an entry point for dataset discovery. At its core, LODVader employs a timely-updated index using a complex cluster of Bloom filters as a fast search index with low memory footprint. This BF cluster is able to efficiently perform analysis on link and dataset similarities based on stored predicate and object information, which -once inverted -can be employed to discover invalid links by displaying the Dark LOD Cloud. By combining all these features, we allow for an upto-date, multi-dimensional LOD cloud analysis, which -to the best of our knowledge -was not possible before.
doi:10.1145/2872518.2890545 dblp:conf/www/NetoMBKH16 fatcat:nontrpfggberxcd2xhjugxnjbm