This paper proposes a system development methodology for a transaction processing system (TPS) based on model theory and shows its feasibility and effectiveness by applying it to an example.

The paper first explains the objective and properties of model theory approach and discusses the reason why a formal approach is needed for a TPS development. The paper next clarifies the model structure of TPS formalized by model theory approach and proposes a development procedure for TPS. TPS is formulated as an automaton composed of both a file system and a process that operates on The process is decomposed into a standardized user interface and a user model, which represents a problem. It is showed that the main task for system development in the approach is to have a set theoretical description of a user model based on the formalized TPS model. Finally, the paper discusses the effectiveness of the approach by showing some system development example.

Each ERP(Enterprise Resource Planning) package vendor, system integrator, and consulting firm has offered different ERP implementation methodologies to support ERP installations for user companies. Many companies have installed enterprise systems using ERP packages with ERP implementation methodology. However, some ERP implementations have ended in failure. One of the reasons for ERP implementation failures is the issues of ERP implementation methodologies, which come from two drawbacks: one is a lack of user-oriented perspective, and another is a gap among those methodologies.

In order to overcome these two drawbacks, it is necessary to develop a seamless ERP implementation methodology from user-oriented perspective. In this paper, we present a framework for "generic ERP implementation project" which includes the user-oriented perspective and important activities.

It is important to perform efficient supply and demand for consumables, which need to be replaced as long as customers who purchased machines keep using the machines. The method was designed to find two types of points by utilizing multiplex accumulation graph; the point where customer demand is actually changing (called " demand changing point") and the point where the demand looks changing but actually not (called "inventory changing point"). Furthermore, the analysis system was developed to identify these points easily. By applying this method and system for 744 actual consumables at the fiscal year-end of 1997, two demand changing points and four inventory changing points were extracted from the entire product life cycle which can be classified into 3 stages such as introductory growth, and decline stages. As a result, demand forecast accuracy, which is one of the big issues in the factory was improved and product inventory was reduced by 30Yo through this improvement. Lastly, the demand forecast model, which supports to make efficient production plan in terms of quantity and timing, was proposed.

This paper deals with the analysis of multi-agent systems on planer cells consisting of local interaction and GP learning and its applications to the analysis of collaboration among firms. As the model of agents' behavior, we assume two types of model. As the first type of agents model, we assume two kinds of agents having own utility functions predict their optimal behavior, and then the market assess the production and employment which is used for the GP learning of agents. As the second model, the single-type agents are assumed to behave on the Prisoner's dilemma game, and their behavior is updated based on the GP learning using prescribed payoff. By simulation studies we show various chaotic phenomena are observed besides the equilibrium. Then, the control method based on GP procedure is proposed which leads the system to the formation of clusters of agents' states. Finally, the model is extended to describe the collaboration and modular production among firms by relaxing the constraints posed on the local interactions.