Development of a stream reasoning benchmark (engineering track)

Abstract

The emergence of stream reasoning accompanies the expansion of the semantic web. A stream is data that is continuously being generated from various sources, such as the stock markets and social networks. The growing number of parallel streams makes it imperative to imitate the rapidly increasing real-time events. The challenges of handling these streams’ time-related aspects call for a reasoner to process various facets of this frequently changing dynamic data. However, the existing stream reasoners are limited in scope. To improve the performance bottlenecks of the current systems and pave the way for further development in this direction, we need to have high-quality benchmarks that can test these reasoners to the limit. To fill the gap of existing benchmarks that are limited in scope, we propose a stream reasoning benchmark that is based on academic Twitter data. The dynamic nature of the Twitter data streams helps test various aspects of the stream reasoners such as scalability, expressivity, handling concurrent queries, data with different frequencies, and parallel streams. We propose this work as a base-line to evaluate the reasoning capabilities and identify the limitations of RDF Stream Processing engines.