|Title||:||A Semantic Approach to RDF Compression|
|Details||:||Tue, 18 Aug, 2015 3:00 PM @ BSB 361|
|Abstract:||:||Over the recent few years, there has been a surge in the number of Linked Open Data (LOD) datasets that are being published by the Semantic Web community in RDF formats. In case of large datasets, publishing them and exchanging them between users in uncompressed form becomes a challenge. Hence, there is a need for compressing such datasets. RDF datasets inherently represent the semantic associations prevalent among the real-world entities. Exploiting such semantic associations can help in achieving better compression. The current state-of-the-art techniques focus on providing compact structural representation of RDF data by taking just the syntactic redundancies into consideration. Also, RDF datasets provide incomplete information and follow open world assumption (OWA). Thus, they can be subject to updates at later stages. Structural representation techniques provide a static dictionary mapping and ordering in adjacency lists that hinder incremental updation of datasets in the compressed form. We observe that mining rules that inherently take OWA of datasets into consideration can be effective for incremental compression.
In our work, we propose a rule-based RDF compression technique that efficiently exploit semantic associations to achieve better compression. We also propose an incremental compression technique that avoids re-compression of the dataset each time there is an update. Our experiments show the effectiveness of the proposed compression and incremental compression techniques.