Detection of problems and missing stakeholders during software development meetings

The project is an ongoing research activity under Grace (Center for Global Research in Advanced Software Science and Engineering) Tokyo, Japan.

Abstract

This project is concerning the exploration of developing a refined method for analyzing software development meetings. We want to provide a method that can detect lack on consensus and missing stakeholders during the meetings. The expected result is a system that provides a simple visualization to project managers that helps them to detect problematic situations and topics, and possible missing stakeholders.

Background of research

The research was motivated by the problems in stakeholder identification in requirement engineering. Stakeholder identification is an activity that takes place mainly during requirement elicitation activity in requirement engineering. The IEEE Standard 830 gives a summary of the properties that should ideally be part of software requirement specification: correct, unambiguous, complete and consistent. Therefore, any identification processes that mistakenly recognize someone as a stakeholder will probably include requirements, which do not correspond to any real need (a feature of ‘Correctness’ of the standard) (Pachero et al, 2007). On the other hand, when the identification task fails to detect participants who are needed (someone is missing) for the software project, requirement specifications are no longer ‘Complete’ due to the omission of relevant requirement specifications success, and this could give rise to inconsistent specifications. Failing to obtain these properties can create risks that could affect the project. This requires a previous stakeholder identification process (Pachero et al, 2007). However there have been little work done and staggering progress in this direction. So how do we solve this problem? We discuss in the following.

Our research proposal

In order to get a better grip of the research problem and how to solve them, we carried out an empirical study. We selected a particular group of animators developers meetings recorded, which have been recorded on FlashMeeting video-conferencing system (in collaboration by the FlashMeeting group at Knowledge Media Institute, Open University, UK). The meetings were used as an benchmark to develop our methods of analysis. We show the user interface of FlashMeeting Replay:

Fig.1: FlashMeeting video-conferencing system.

After 3 months of analysis, we found that it is inherently difficult to identify missing stakeholders in an animation meeting. The reasons were that the meeting itself was ill-structured and random. Hence, we suggest that in order to solve the problem requires an understanding at first- at a contextualized level about:

(i) who has said what (topics),

(ii) in which situation where problems pertain or might arise,

(iii) who is doing what (functional role), before systematically identifying who should be there to say more about what (i.e., who is missing here)?

Therefore the goal of our study is to systematically investigate how a framework can be conceived that can explicate knowledge about topics, situations, and meeting phase. And then from these levels of knowledge, how we can systematically organize the details to infer who is missing. Our aim is to develop a simple yet efficient visualization that can detect problematic situations, topics enabling the detection of missing stakeholders.

Our research results (so far and ongoing...)

The input to our work are annotated communication exchanges, both audio and chat-based messages. The audio was transcribed professionally. The work involved trying to align the audio and chat-based messages to make sense of what is taking place. The output of our results is illustrated below:

Fig.2: The output from conducting the empirical study.

The figure summarizes the resulting output from our empirical study. From the observations of the meeting, we developed a communication analysis model to help us tie the annotated communications, whereas the emergent coding helps to organize the annotated communications into levels of information. The simple visualization is shown below:

Fig.3. Detection of problematic situations, and topics.

In Fig.3 we illustrate a marked up design of a simple visualization for detecting problematic situations, and topics. For example the action, express concern can indicate potential problems and content:focus; silhouette: humbled up indicates what is the problem.

Remaining challenges

It remains a challenge to develop an analysis that can extract the appropriate contents that is very important for project managers and requirement engineers. Specifically in selecting what is potentially the focus of discussion, and then allowing an efficient visualization of who is missing. Another challenge is how do we design an efficient visualization that shows detection of: (i) each developers problem(s) and group problems, and (ii) missing stakeholders through tracking problematic topic?

Research plans

Our research plans are the following is to incorporate several rules:

1. Rules for content selection (knowing what to eliminate as content is very important for project managers).

2. Rules for selection of topic in association to content.

3. Rules for detection and tracking of problematic topics, not discussed or solved.

4. Rules for detecting missing stakeholders

4. Refine the visualization that allows efficient detection of missing stakeholders.

Finally our plan is to apply the method and visualization to a practical meeting in industry and getting feedback.

Collaborators:

Prof Shinichi Honiden, professor and director for Information Systems Architecture Research Division, National Institute of Informatics, Japan.

Dr. Eric Platon, visiting research at National Institute of Informatics, Japan.

Dr. Toshihiko Tsumaki, requirement engineer at Nihon Unisys Ltd, Japan.

Prof Bashar Nuseibeh, research director for the Computing Department, Open University, UK.