Identifying Similarity of Software in Apache Ecosystem - An Exploratory Study

Majority of software depends on external libraries to achieve their goals. We conducted studies to identify software similarity regarding libraries; using 91 Java open-source projects managed by Maven from Apache ecosystem and considering the Apache Software Foundation (ASF) categories for every project. We extracted software libraries to compute the similarity thought two similarity metrics: Jaccard and Cosine. Results showed one software without external library and low similarity degrees among software libraries caused by a highly disproportion quantity of dependencies per project; for instance, three projects had 100% similarity caused by depending just on JUnit library, which is also the most used dependency on analyzed projects. Although our study revealed a low similarity among selected projects, it provides useful information about trending libraries for developers based on the relation among most used libraries in Apache projects, regarding ASF categories and Apache community preferences. It also discloses reasons for similarity degrees and for most used dependencies on Apache ecosystem.