Due to the increased computation capacity of personal computers, the everyday use of expert systems is almost within reach for the non-professional users as well. The results of this development will offer huge help in making decisions in situations which require complex expert knowledge.
Aforementioned situations might include financial decision-making that has already become a daily routine since the increase of on-line commerce (e. g. with the development of on-line marketplaces), but these decisions are still below the complexity of those required in the corporate world. There are quite a few major differences between these two fields of use, like the number of transactions, the variety of goods traded, the applicable laws and obligations and the knowledge and experience of the participants. Generally speaking, a system created for individual users must adapt to the conditions of a very different market environment where the available data is poorly defined and the appropriate statistics are unavailable.
These particular requirements increase the significance of using alternative approach such as the fuzzy method. Fuzzy methods are suited for either managing uncertainty in data or representing evolved knowledge, or, in addition to that, to effectively reduce model complexity. There is, however, a serious disadvantage to that: the fuzzy sets that represent the conclusions are basically static, thus they don’t produce adequate results in a dynamically changing environment. This disadvantage is usually balanced by the available assets of machine learning: the fuzzy sets that are the basis of the conclusions are generated automatically, allowing the adaptation of the system in the same time.
The goal of the thesis is to establish a feasible system architecture and evaluate the available possibilities through the creation and simulation of a system designed for demonstrational purposes within the aforementioned small scale economic environment. The system model incorporates the collection, systematization and interpretation of data that is required for decision-making, and it also involves the frameworks of the decision-making itself as well. This framework primarily applies fuzzy solutions but also utilizes some elements of classical financial analysis.