For planning and enforcing the On Line Maintenance (OLM) in the near future, we should take care of keeping a reasonable balance between the degree of risk increase and the task volume equalization. In this paper, the current status of studying for the OLM introducing on Japanese utilities is described. The planning approach of OLM is examined on the assumption that planned OLM can be carried out. The practical methods and evaluations for the OLM are described, and the trial evaluation for this method is also described.

Recent years have seen a shortening of product lifecycles and a trend toward product optimization for individual customers, but at the same time, development and maintenance departments have become increasingly global in scope making it difficult to improve product quality in the market. Against this background, the need has been felt for a system that can make efficient use of failure information knowledge accumulated for different products and by different departments to improve maintenance response and prevent reoccurring failures. Although advances in information processing technology have driven the evolution of practical information -storage and information -retrieval systems, the search performance of an interoperable system handling failure information knowledge across different products and departments suffers from many problems in terms of effectiveness and efficiency if using the standard keyword search method. This paper proposes a new information retrieval system focusing on information architecture differences between departments and presents the results of evaluating an actual failure information system using this system.

Many industrial organizations are interested in using the past failure experiences for the educational purposes for the prevention of future troubles and accidents as well as the improvement of the quality of their products and services. The education often employs the style of drawing lessons from similar experiences in the past, and, concerning the similarity, it is important to look at the background causes and tacit factors such as the organizational culture and value systems rather than the superficial phenomena such as wrong operations. However, the analysis of such background factors tends to depend on human subjective judgments and personal impressions. Aware of such problem, this paper discusses an approach to establish a framework to give the guideline for the analysis. Referring to Y. Engestroem’s activity theory, our framework describes a dynamical process where a small seed of a problem propagates and grows to an explicit problem such as a trouble or an accident. The analysis is a process of interpreting the facts in terms of the activity structure, and this enables the generalization and similarity evaluation of an individual case in terms of the activity structure, beyond the superficial and domain-specific descriptions. Using a example of a quality defect trouble in a manufacturing process, we explain our approach and shows how an accident in a different domain can be drawn as a similar case by looking at the inner activity structure.

How to facilitate the process of consensus building among the partners in a supply chain is a crucial issue. As an example of such a problem, this paper takes up perceptual mapping and develops a prediction market system for collectively determining the location of an object on a two-dimensional perceptual map. It introduces a new type of prediction security called SAPS (Square Area Prediction Security) and extends a widely used market making algorithm LMSR (Logarithmic Market Scoring Rule) for handling SAPS in a batch updating environment. This system feeds the location of the object back to the participants as the price distribution of SAPS, which is determined by the number of SAPS sold so far in each square area, and allows them to modify it through transactions of SAPS. The paper further conducts laboratory experiments to see whether the proposed system can support a group of people to build consensus on a perceptual map. As a result, it is observed that the variance of their subjective evaluations on the location of a specified object is reduced through operating the system but that the evolving consensus is not always captured properly by the price distribution. What causes this problem and how to fix it are also discussed.