Distributing Computationally Expensive Matching of Requirements to Capability Models

In this paper, we present a distributed way to automatically map users' requirements to reference process models. In a prior paper [9], we presented a tool called Process Model Requirements Gap Analyzer (ProcGap), which combines natural language processing, information retrieval, and semantic reasoning to automatically match and map textual requirements to domain-specific process models. Although the tool proved beneficial to users in reusing prior knowledge, by making it easy to use process models, the tool has one main drawback. It takes a long time to compare a very large requirements document, one that has a few thousand requirements, to a process model hierarchy with a few thousand capabilities. In this paper, we present how we solved this problem using Apache Hadoop. Apache Hadoop allows ProcGap to distribute matching task across several machines, increasing the tool's performance and usability. We present the performance comparison of running ProcGap on a single-machine, and our distributed version.