Research On Qualitative Simulation For Knowledge Group Behavior In Enterprises

The uncertainty of environment for enterprise has been intensified by the internationalization of market. In order to enhance the flexibility and reaction speed of enterprise, many groups or teams are set up in enterprises. In group, innovation ideas of individual could be blended to enhance the creativity of enterprise. So many professionals have been attracted to work in this field. In these studies, common research methods include experimental verification and demonstrational analysis. But modern auxiliary techniques such as computer simulation have been rarely taken. The advent of the qualitative simulation provides a new way for the research of group behavior. The characteristics of cooperation in knowledge group are analyzed firstly, in which its non-zero game characteristic is discussed. In most researchs, only two choices are concerned: cooperation or no cooperation. In this dissertation, the degree of cooperation is concerned. Sub-groups'income is fuzzified to make it more close to the fact. Therefore, a hybrid qualitative modeling method is designed. For the known quantitative relations within variables, mathematical function, i.e., quantitative model, is built. For the known qualitative relations, the logical relation, which seem as qualitative "+"and "-", is built. The trends of change for system are simulated by integrating the two models with common variables, and using auxiliary knowledge. A gaming model of two-person team is then built, in which the rules, algorithm and steps for simulation are designed. Group consists of different sub-groups. The interactions within them are neither linear nor hierarchical in the formal organization. Some sub-groups may be the leader of the group, their behaviors may take great influence on the others. So an integrated qualitative causal reasoning algorithm is designed to describe and explain behavior changes of different sub-groups. The description of variables, distinctive time points, distinctive value and causality of group behavior are interpreted. A causality graph of group behavior is depicted. Then the method of combining QSIM algorithm with basic causal reasoning technology is studied. Representations for interaction within management strategy, complexity of environment and group behavior are designed. Simulation rules and simulation mechanism are given. Finally, this method is applied to an example of management decision on motivating employees. The validation of the method is taken. The characteristics of task-oriented-group are analyzed. Then a work process model for task-oriented-group is designed according to the method of cell decomposition. In this model, the work process of group and environment of group are taken as key points. Rules for the interaction of variables are designed. Common sense and expert knowledge are used to constrain transition of variable state. Probability distribution is adopted to resolve the uncertainty of transition. The behaviors of system are yielded by simulation of many times other than one time. Finally, nine situations are chosen to show application of simulation in term of different cross combinations of leader characteristics, management strategy, interference in interpersonal relationship and interference in work. Simulation results indicate that, combination explosion could be avoided by two ways. First, the rules are set up by combining common sense with expert knowledge; Second, probability distribution is used to constrain state transitions. The characteristics of behavior change of group can be simulated using our method. Based on the above researches, the prototype of qualitative simulation system for task-oriented-group behavior is developed. The structure and functions of the system are builded. The modules including initial setting, knowledge management, simulation control, strategy evaluation, result management, are designed. Object oriented programming and qualitative simulation technology are integrated. The qualitative simulation module and other relevant functions are also coded by VB6.0.