Applying mapreduce to spreading activation algorithm on large rdf graphs

Abstract

Over the recent years, the Semantic Web has experienced a considerable growth. Governments and organizations are putting major efforts in making information publicly available using Semantic Web formats. Algorithms such as spreading activation have effectively been used for finding relevant and related information on Semantic Web datasets. But, as the Semantic Web grows, these datasets quickly outgrow the computational capacity of a single machine. The same computational problems found in the past in the traditional web arise. On the other hand, computational frameworks like MapReduce have proven successful resolving problems that handle large amounts of data. We introduce an implementation of the spreading activation algorithm using MapReduce paradigm, discussing the problems of applying this paradigm to graph problems and proposing solutions. Hereby, a concrete experiment with real data is presented to illustrate the algorithm performance and scalability. © Springer-Verlag Berlin Heidelberg 2013.