Natural Language Processing (NLP) based Extraction of Tacit Knowledge from Written Communication during Software Development

Abstract

Software engineering, in general, and Global Software Engineering (GSE), in particular, face challenges such as handling communication and collaboration issues and inappropriate knowledge management. It is difficult to ensure availability of right knowledge at the right time to the right person during software engineering activity. The knowledge needs to be shared across the organization but limitations of knowledge sharing tools or dispersed knowledge sharing media and improper handling of tacit knowledge makes it more challenging to share knowledge. Significant amount of communication during software engineering process takes place via emails and discussion forums. Decisions taken during planning activities are also communicated through emails. These emails and the discussions do not necessarily cover the rationale behind a decision or approach adopted. This leaves a substantial portion of shared knowledge as tacit. Mostly, due to undefined tacit knowledge handling procedures, organizations suffer with the loss of critical knowledge and information. One way to handle this loss of knowledge is to store every information in knowledge repository but this can increase the size of the knowledge repository. This paper presents a three step approach to identify tacit knowledge in the written communication such that the organizations can save the tacit knowledge for future use. The presented approach does not only extract up to 57% of tacit knowledge but also indicates that only 20% of the whole communication need to be externalized hence saving 80% capacity of the knowledge repository, if every communication was to be preserved as potential tacit knowledge.