【作者】 吴颖敏；

【导师】 蔡淑琴；

【作者基本信息】 华中科技大学 ， 管理科学与工程， 2009， 博士

【摘要】 全球经济使得企业的产品或服务易被模仿且进入门槛更低，要求企业在市场动态环境中需要快速响应、缩短创新周期，发现并把握新的市场机遇成为企业决策者关注的焦点和工作重点，构建企业市场机遇发现的支持系统成为企业信息化的深度需求。本文在综述国内外机遇发现的理论研究和支持方法的基础上，针对已有研究中尚待解决的非精确半结构化数据源、复杂关系、柔性结构、形象思维、过程支持等问题，运用认知科学分析机遇发现的机理，基于超图理论提出新的决策支持模式和方法。超图是普通图的推广，具有较普通图更强的建模能力以处理非精确、不确定性。本文首先介绍了超图的基本概念和图形表示；指出超图的建模特点、应用领域，对已有的超图模型进行了综述；对超图的发展脉络进行了梳理，着重介绍了超图在数据库领域和知识建模中与其他模型的相互关系和相对优势。针对非精确、不确定、半结构化环境中的机遇发现，运用认知科学中的图式理论分析了机遇发现的机理，从认知的角度重新界定机遇和机遇发现；基于超图理论对主体的相关图式建模，提出了机遇发现的超图系统模型；设计了基于超图系统模型的支持机制，支持机遇发现过程中的数据重组、模式识别、路径分析、信息吸取、图式构建，以及可视化呈现和交流。模式识别支持是感知、联想等形象思维的基础；机遇发现不仅有对事物属性的识别，更主要的是基于原型或图式所进行的模式识别；通过定义超图系统模型中的各类模式基元，运用KDD中的频繁模式挖掘算法提取非精确、半结构化的信息源的模式基元，基于模式基元描述了图式的立体结构：分析了机遇发现中的两种模式识别，包括基于属性的模式识别和基于结构的模式识别。针对后者提出了基于超图不变量和特征向量法的同构判定方法。路径分析是支持机遇搜索、评价、解释的基础；定义了路径分析在机遇发现中的特殊语义，即一条超路径描述了一个问题的可行解；分析了已有超路径求解算法的不足；提出了基于两点的超路径求解算法，解决了指定两实体之间的可行解求解；提出了基于起始点的超路径求解算法，解决了给定实体的所有可达解域。信息吸取是信息交互和图式构建的基础，体现了超图系统模型对机遇发现动态特性的支持；研究了机遇发现作为开放的认知系统，与外界环境的信息交互，并由此导致的图式系统演化，即信息吸取；详细阐述了在信息吸取中的几类关系运算算子；针对同化的信息吸取，设计了以自然连接运算为主的数据库操作；针对顺应的信息吸取，设计了基于遗传算法的图式构建和创新。本文以商业银行应用领域为背景，建立商业银行机遇发现的超图系统模型；基于超图系统模型，分别给予模式识别、路径分析、信息吸取的决策支持；对比已有机遇发现支持技术和方法，验证了超图系统模型的支持特点和优势。更多还原

【Abstract】 The global economy makes business products or services can be easily imitated and lower barriers to entry, thus the enterprises in the dynamic market environment are required to response rapidly and shorten the innovation cycle. Finding and seizing new market opportunity has become the focus and emphasis of business decision-makers, and building support system of enterprise market opportunity discovery become the deep needs of enterprise informationization.This article is found in the summary of both domestic and international theory research and support methods, for the issues found in current study that have yet to be resolved such as non-accurate, semi-structured data sources, comlicatied relationship, flexible structure, the image of thinking, the process supporting and so on, using cognitive science analyzed the mechanism of opportunity discovery, raising new decision support models and methods based on the hypergraph theory.The basic concepts and graphics express of hypergraph are introduced. The modeling features and application fields of the hypergraph are pointed out, the current hypergraph models has been reviewed, The development network has been arranged, and the mutual relations and comparative advantages between hypergraph and other models in the application of database fields and knowledge modeling are introduced in emphasis.For opportunity discovery in non-accurate, uncertain and semi-structured environment, its mechanism is analyzed by using schema theory of cognitive science, opportunity and opportunity discovery are redefined from a cognitive point of view. Hypergraph system model of opportunity discovery is raised based on relative schema modeling to agent in hypergraph theory. Support mechanism based on hypergraph system model is designed to support data reorganization, pattern recognition, path analysis, information absorbing, schema building, as well as visualization and communication during opportunity discovery process. Comparative features and advantages of hypergraph system model to current model are analyzed.Pattern recognition being the basis of image thinking such as perceived association is clarified. Pattern recognition in opportunity discovery process is studied. various types of model primitives of hypergraph system model are defined. Frequent patterns mining algorithm in KDD is used to extract model primitives from non-accurate and semi-structured information sources, and Three-dimensional structure of schema is described based on model primitives. Two types pattern recognition in opportunity discovery is analyzed, one is based on properties, the other is based on structures. For the latter one, isomorphism determining method based on hypergraph invariants and eigenvectors is proposed.Path analysis being the basis to support opportunity for search, evaluation and interpretation is clarified. Special semantic of path analysis in opportunity discovery is defined, which means a super-path describing a feasible solution of the problem. Inadequacy of the current ultra-path algorithm is analyzed, while a super-path algorithm based on two points is proposed to solve problem of a given entity in all solution domain.Information assimilating being the basis of information exchange and schema construction is clarified, which reflects the support of hypergraph system model to the dynamic characteristics of opportunity discovery. Information assimilating, that is, as an open cognitive system, opportunity discovery interchange information with external environment, and result in schema system evolution, is special studied. Several types of relational operator in the information assimilating are described in detail. A database operation mainly based on natural connect computing is designed for Assimilate information assimilating. Schema construction and innovation based on genetic algorithm is designed for acclimation.Applications to commercial banks as the background, the hypergraph system model of commercial banks opportunity discovery are established. Based on hypergraph system model, the decision support for pattern recognition, path analysis and information absorbing is given. Compared with the current support technology and methods of opportunity discovery, the features and advantages of hypergraph system model are validated.更多还原