18-09-2012, 02:51 PM
Oriented Knowledge-Management System
Oriented Knowledge.docx (Size: 33.8 KB / Downloads: 24)
Abstract:
We have developed a model of a review-report-oriented knowledge spiral that aims at guiding
projects to their goals steadily as well as promoting knowledge reuse in software projects. In
order to attain these aims, we focused on reports of design review. The features of our model are
(1) to manage the discussion processes in design review, as well as documents and program files,
(2) to store the discussion processes written in the review reports and to share them in an
organization, and (3) to provide evaluation standards to help a manager manage the discussion
processes. We also developed a WWW-based knowledge-management system based on our
model, and it is being applied in more than 50 projects. And the system has been qualitatively
shown to improve the quality of software development. We also show a measure of quantitative
evaluation for our proposal.
Introduction
With the dramatic advances in Internet/Intranet technologies, the number of distributed software development
projects has been increasing. And Internet and Intranet groupworking tools are becoming more widespread [1]
and can often provide a low-cost, easy to implement solution to share and utilize knowledge in world-wide
organization. On the other hand, in addition to the conventional technology of software components and its reuse
technology around CASE tools, the research on knowledge acquisition, retention, and reuse in a software project
has been more active. Maurer has been trying to improve the software development efficiency by putting the
development process on workflow management system and presenting the necessary knowledge in each task to
the members[2]. Conklin structuralized discussions in the upper stream of software development and proposed the
IBIS (Issue Based Information Structure) model which enables understanding of the discussion context by
visualizing the discussion process[4]. Komiya improved the IBIS model to make software design more efficient
by recording and reusing the discussion process under a distributed development environment[3].
We assume that in order to improve software quality, discussion during design reviews must be managed. We
thus developed a knowledge spiral model for utilizing recorded discussions. Then we developed a Web-based
knowledge management system based on this model. And this system is being used in more than 50 software
projects. In this paper, firstly we clarify the knowledge source in each phase of the software development process
Proposal of Review-Report-Oriented Knowledge-Spiral Model
Development Process as Knowledge Creating Process
Software development is essentially a knowledge-creating process. Designers should define problems clearly,
gather information to solve them, and choose proper solutions or create new ones. This process is the activity to
keep producing knowledge toward the final goal. We use “problems” here as the items for assurance of software
quality and feasibility. The designers lay the foundation for new knowledge creation by using past development
experience or technology that has been piled in the organization.
Conventional software-development models often measure progress according to the extent of artifact
registration. This approach, however, does not guarantee that the design will be fully discussed.
From Artifact Management to Discussion Management
Nonaka modeled the knowledge-creating process as a tacit and explicit conversion process. This knowledge
conversion is called a “Knowledge Spiral” because new knowledge is piled upon existing knowledge[6]. When
Nonaka’s knowledge spiral is adopted as a knowledge-creating process in a software project, it is important that
an organization provides the discussion place and rotates the knowledge spiral in order to improve development
efficiency. In other words, project members need a process model to encourage knowledge sharing and
knowledge creation.
In light of the above-mentioned requirement, we have added a mode for checking the discussion condition and
have developed our original knowledge spiral model. And we have solved the problems that occur in
conventional software-development-process models on the hypothesis that the quality and quantity of discussion
rather than those of artifacts should be managed.