A Study On Case-based Decision Support Techniques And System For Opportunity Discovery

With the rapid development of global economy and booming science and technology, the dynamic changing of market environment leads to increasing uncertainty of business transactions, and enterprises are facing increasingly fierce market competitions. Enterprises that can not effectively discover and seize market opportunities will face the risk of being eliminated at any time. Especially in the context of the financial crisis, opportunity discovery has become the hotspot and focus attended by industry and academia. However, in the market, there emerge more and more the sources and the amount of information, the complexity and concealment of which is increasingly high, and the time frame for enterprises to acquire information, analyze, then make decision is becoming shorter and shorter. Consequently, it is difficult to discover opportunities with former artificial methods. On this occasion, it is urgent need for survival and development of enterprises to utilize information technology to build opportunity discovery support system, supporting decision-makers to make better use of the available information and knowledge to discovery opportunities in the market.Since Ohsawa, the Japanese scholar proposed Chance Discovery (CD) issue in 2000, opportunity discovery has been receiving more and more attentions by domestic and foreign scholars. A number of theoretical models and supporting techniques are put forward, among which there are still some spaces for improvement in aspects of theoretical methods, implementation techniques and applications systems. In tracking the development trend of opportunity discovery, decision support, artificial intelligence and other related subjects home and abroad, integrating the latest research achievement in multi-disciplinary fields, and taking qualitative and quantitative analysis methods as acombination, this dissertation studies the aspects of intrinsic mechanisms, essential features, decision processes and support techniques of opportunity discovery as follows:Firstly, the dynamic characteristics of opportunity discovery are analyzed from internal recognition external environment. A three-dimension space is constructed with the abstracted three dimensions of demand, supply and operation, in the framework of which the root schema and evolution mechanism of opportunity discovery are investigated. By analyzing the characteristics of opportunity discovery decision, a decision support framework of opportunity discovery based on CBR is proposed.Secondly, aiming at the case representation issue, opportunity discovery tree model is proposed to describe opportunity discovery process. On studying the information hierarchy and multivariant relationship of opportunity discovery, hypergraph system for opportunity discovery tree is put forward based on the concept of information system and hypergraph. Then the granularity description and calculation method of hypergraph system is studied.Thirdly, towards case retrieval issue, existing similarity measuring metheds are summarized, and a model of similarity metrics for hypergraph system is proposed, the computational procedure of which is consist of comparability analysis, correspondence solving, and similarity degree computing. For the correspondence solving of vertices in hypergraph system, a maximal matching enumeration algorithm for bipartite graph based on comparability adjacency matrix is proposed. The usefulness of our model and algorithm is verified by two groups of experiments.Forthly, directing at learning of opportunity discovery cases, the issue of opportunity discovery activity pattern mining is presented. By weighting activities concerning different importances, a opportunity discovery activity pattern mining algorithm based on weighted support, wGSP algorithm, is proposed. On the results of activity pattern mining, a similarity matching method for opportunity discovery activity sequences is studied.Fifthly, according to problem solving theories and methods, the problem solving based opportunity discovery decision support process is studied, including the recursion of information assimilation, case retrieval and case revision stages. The ways, operators and methods of information assimilation, factor model and mode of case retrieval, ways, operators and methods of case revision are studied respectively. A heuristic algorithm for case revision based on k-star hypergraph extraction is proposed. The working process of case-based opportunity discovery decision support is elaborated.Sixthly, the working requirements of decision support and working mechanism of DSS for opportunity discovery is analyzed. By introducing Multi-Agent technique, the structure and working process of constituted Agents are investigated, and the cooperation model of Multi-Agent for opportunity discovery decision support is proposed. The architecture and functions of case-based opportunity discovery decision support system are presented, and the platform and an example of system implementation is shown.Finally, the contents and conclusions of this dissertation is summarized, and the perspective for future work is presented.