The field of software engineering embraces measurement, analysis and modeling of software. Software metrics are often based on counting, whereas this thesis adopts information theory. The goal of this research is to show that information theory-based metrics proposed by Allen can be useful for software development projects compared to counting-based metrics. Briand, et.al. have defined five families of measures based on counting the elements of a graph. This research considers a hypergraph system. Parallel Mathematical Library Project (PMLP) was used as the case study. Abstract semantic graphs were generated for the C++ source files of PMLP in the form of nodes * hyperedges tables, which are measured for counting and information theory-based measures. Analysis showed that information theory-based metrics provide fine-grained distinctions among the modules, compared to the counting-based metrics. The case study measurements conformed to the properties proposed by Briand et.al. as well.