Mahasweta Chakraborti, University of California, Davis
Open Source Software(OSS) projects are driven by communities of independent yet dedicated developers. Our work explores the forms of Institutional statements(rules/norms/strategies) core to large OSS projects under the mentorship of the Apache Software Foundation. We quantify Institutional statements in the OSS context, and develop automated means to recognize the same across email correspondences between project members, users of the software product as well designated overseers or Apache itself.
Natural conversations between developers do not always appear as formal, structured exchanges that can be interpreted along the A(B)DICO syntax, but are instead often dispersed through an email or presented through informally phrased discussion. This limits identification through expression matching or similar approaches. A challenge also emerges in discerning the scope of “rules” exchanged, especially whether they are relevant to Institutional Analysis and the community, or are simply every day ground-level directives between specialized contributors towards product development and operations. All such rules, institutional or not, feature extremely specific, technical exchanges and thus present semantic similarity. We report our findings on understanding the nature of Institutional Statements in OSS projects, data collection, and their eventual automated extraction from email exchanges using NLP methods grounded in language modeling. We shall also discuss the implications of such analysis on a project’s progress and prospects through the incubator program.
