Auto-tagging Emails with User Stories Using Project Context

Abstract

Abstract. In distributed agile teams, people often use email as a knowledge sharing tool to clarify the project requirements (aka user stories). Knowledge about the project included in these emails is easily lost when recipients leave the project or delete emails for various reasons. However, the knowledge contained in the emails may be needed for useful purposes, such as, re-engineering a software, changing vendor and so on. But, it is difficult to relate texts such as emails to certain topics because the relation is not explicit. In this paper, we present and evaluate a technique for automatically relating emails with user stories based on their text and context similarity. Agile project management tools can use this technique to automatically build a knowledge base that is otherwise costly to produce and maintain.

Review of approaches to manage architectural knowledge in Agile Global Software Development

“…Another source of knowledge are email messages, but since they are unformatted (free text), they must be classified or categorised. In [44], a machine learning technique is applied to automatically create relations between emails and user stories. This technique assigns a tag (representing a user story) to each message, thus when a developer finds a message, s/he knows the related user story.…”

“…However, frequently there is an insufficient AK in wikis of open source projects [31]. A similar situation occurs in AGSD projects, so people are forced to capture AK to keep the project documented [44]. Despite the reported problems, the adoption of wikis might reduce the time to find AK [31, 61].…”

“…Also, since time pressures are common in Agile environments, auto‐generated artefacts seem as a good option to codify AK, as they require a lower effort from developers. We found only four papers about this topic, which present solutions to show system structures [47], to generate experts’ directories [76] (as in Bjørnson and Dingsøyr [22]), to trace artefacts [45, 46], and to relate email messages with user stories [44]. We consider that a new path to explore is automatic artefact generation from repositories commit messages or from logs of electronic textual media, since these are everyday tools for developers.…”

“…Reported impact/benefit software development support [43] -Users were more inclined to have daily meetings to review the overall architecture of the project supported with Archinotes. auto-generated [44] -The matching accuracy between email and user story of the auto-tagging mechanism was 70%. [47] -WRCC was useful to speed up the product's description to new team members, to visualise product's complexity, and to understand architectural changes' impact.…”

Towards a reduction in architectural knowledge vaporization during agile global software development

“…Despite the above problems, literature reports that developers prefer using free tags because of their low cognitive load for everyday work [44,45] . Some efforts have been made to develop auto-tagging mechanisms [46] or tag-based recommender systems [43,47] to reduce developers' cognitive loads to an even greater extent. Sohan et al [46] auto-tagging mechanism relates email messages to user stories in ASD projects with an accuracy of 70%; however, the remaining 30% of error could cause knowledge retrieval problems.…”