Slack's Knowledge Classification Mechanism for Architectural Knowledge Condensation

摘要

Agile methods adoption in global software development (GSD) has revealed architectural knowledge (AK) management as a challenge, since documentation debt prevails in these environments. Thus, knowledge dissemination/acquisition becomes difficult because of the four inherent distances in GSD, which, in turn causes AK vaporization. One way to reduce this loss of knowledge is to take advantage of the AK traces that different unstructured and textual electronic means (UTEM) leave (e.g. instant messengers and email), which developers prefer on AGSD environments. These traces are valuable, but their access is difficult because these means have no tools adapted to search AK and because this knowledge is scattered in different UTEM. The AK condensation concept proposes that trough a mechanism of classification and collection of messages contained in UTEM, AK could become available for the stakeholders. This article presents a Slack complement, which implements an AK classification mechanism based on social tagging. We evaluated this mechanism with undergraduate students through a replica of a previous study. We obtained positive results in regards to usefulness and ease of use, but with tagging behavior results that differ (not statistically significant) from those obtained previously. These evidences encourage us to evaluate the AK condensation concept in its whole, in real scenarios.