Expressive Rule Pattern Based Compression with Ranking in Horn Rules on RDF Style KB

With the increasing growth of linked open datasets published in the RDF format, demands for RDF compression has been becoming extremely important. There are a number of structured compression schemes that consider the structural redundancies. A very few researches have done that focuses on compact representation. In this study, we compress the RDF datasets by using different semantic association rules learned from the RDF graphs. From the RDF datasets, our proposed system mines the logically related Horn rules for accomplishing higher compression. The system keeps the triples that match the antecedent part and deletes them while match with the head part of the rules. We have proposed the grammar based pattern system (GBS) to reduce the search space size and faster projection. Moreover, we also have proposed the ranking rules and top-k approach for effectively utilizing all the rules and decreasing the mining time. The experimental result affirms that our proposed system has achieved greater compression compared to the existing RB compression approach.