Russia, Rostov-on-Don, Rostov State Academy of Economy

In the our article we offer the decision support system (DSS) "EIS-Manager" which uses the neuro-fuzzy technology. It addressed to help managers in the knowledge management processes that arose a critical factor of the intellectual advantage. DSS can look for the crisis situation and extract the significant patterns from the data.

Россия, г. Ростов-на-Дону, Ростовская государственная экономическая академия

В статье мы предлагаем систему поддержки принятия решений "EIS-Manager", которая использует нечетко-нейронную технологию. Она предназначена для того, чтобы помощь менеджерам в управлении знаниями, которые стали критическим фактором интеллектуального преимущества. Система может выявлять ситуации кризиса и извлекать значимые модели из данных.

Nowadays the business deal with growing complexity and uncertainty. A lot of the complex questions arise. How should the firms change themselves to better adapt to new environmental conditions and change those conditions. How can they to ensure organizational sustainability in a complex, uncertain and interconnected world. The well known scientific methods decide organizational problems are required formalizing and having perfect information. As results, traditional methods can not taking into account a number the factors with the uncertain and qualitative nature (1). The Artifical Intelligence achievements and powerful computers change this situation. The neural net's models are able to learn from the experience. And fuzzy logic theory is enabled using a linguistic models having a powerful describing capability. Combining the well known formal models with the new neural and fuzzy models we can to get the tool which will be more adequate an organizational practice as a complex adaptive system (CAS).

In the our article we offer the decision support system (DSS) "EIS-Manager" which uses the neuro-fuzzy technology. It addressed to help managers in the knowledge management processes that arose a critical factor of the intellectual advantage (2).

Features of our system is in two levels of its structure and neuro-fuzzy tecnology. First level is addressed to help managers in the issues business processes efficiency. It has some functional modules that are required the own knowledge bases. Each of it can be understood as the separate fuzzy expert system. The first one has the expert defined diagnostic fuzzy rules and criterions associated with the BPR functional models. So, expert system can look for the crisis situation (3). Once the crisis event is detected the second one expert system is launched. Its function is in finding out the crisis cause. At this point, manager can play with the model thus be able to find out desirable decision and to gain more knowledge. The last subsystem of the first level intended for the data mining. This feature gives the DSS ability to extract significant patterns from the data. We used the multi-layered feedforward neuro-net with the backpropagation algorithm as main tool of data mining. Such decision allows using the noised and short term row data that more adequate the business practice. Extracted the significant patterns can be added in the fuzzy knowledge base as FAM (fuzzy associated memory) (4). As well, neural-net's patterns can be translated into a polynomial regression structure which easier may be interpreted.

Second level named adaptive to help managers in the issues how they should change the organization in response to the changing environment. As mentioned above, at this level decision maker creates models that describe an internal and external environment of the firm. The external models may be scenarios, marketing models etc. Anyway, each of it may be represented as subset of the fuzzy linguistic models. The internal models describe the business process of the firm including the supply chains. We use DEFx specifications as a standart form representation of the functional, information and event models.

As second level is adaptive it include the monitoring function that allows to estimate fitness the firm to the its environment. The "adaptation curves" to give to the managers ability to see how successful -- how fit -- the firm in serving a real needs or wants. We offer to represent adaptation curves as a fuzzy set the consumer needs with the membership function is depending on its satisfaction.

References