本文關鍵詞： 粗糙推理空間 粗糙數(shù)據(jù)推理 上近似 樹型推理空間 內(nèi)涵精度 數(shù)據(jù)關聯(lián)　出處：《北京交通大學》2017年博士論文　論文類型：學位論文

【摘要】：信息科學的研究涉及數(shù)據(jù)處理的各個方面,相關的工作促進了方向的產(chǎn)生,成果的出現(xiàn)推進了學科的發(fā)展。作為信息科學的研究課題或研究方向,數(shù)據(jù)分類、數(shù)據(jù)約簡、數(shù)據(jù)倉儲、數(shù)據(jù)篩選、數(shù)據(jù)挖掘、數(shù)據(jù)推演等針對數(shù)據(jù)處理的課題既表明了研究領域的寬泛與活躍,也蘊含了理論和應用相結(jié)合的研究理念。不同的工作雖各有側(cè)重,但常常涉及共同的研究層面。就數(shù)據(jù)問題而言,不明確、非確定、似存在或潛存于數(shù)據(jù)之間的數(shù)據(jù)聯(lián)系與這些方向無不相關,同時又在實際當中頻繁出現(xiàn),從而引出了粗糙數(shù)據(jù)聯(lián)系的概念。對此的思考和關注促成了粗糙數(shù)據(jù)推理課題的產(chǎn)生,較少的涉足預示著研究的意義和前沿,加之理論研究將提供算法構(gòu)建的依據(jù)以及程序設計的基礎。因此本文聚焦于粗糙數(shù)據(jù)推理課題的研究,完成的工作集中于如下幾個方面:對粗糙集依托的近似空間進行了結(jié)構(gòu)上的擴充,引入了推理關系,產(chǎn)生了粗糙數(shù)據(jù)推理得以實施的依托環(huán)境—粗糙推理空間。為對粗糙數(shù)據(jù)聯(lián)系進行描述,在粗糙推理空間中,通過等價關系與推理關系融合信息的上近似,引出了粗糙數(shù)據(jù)推理的定義,使推理運作于數(shù)據(jù)之間,產(chǎn)生了課題研究的主題。經(jīng)對粗糙數(shù)據(jù)推理的研究,獲得了相關的結(jié)論,展示了粗糙數(shù)據(jù)推理的性質(zhì),包括:粗糙數(shù)據(jù)推理保持確定數(shù)據(jù)聯(lián)系的特性,粗糙數(shù)據(jù)推理與上近似中近似信息密切相關的特性,粗糙數(shù)據(jù)推理具有近似描述功能的特性,粗糙數(shù)據(jù)推理與路徑相互等價的特性,粗糙數(shù)據(jù)推理對應不同等價關系的特性等。構(gòu)建了實際問題的粗糙推理空間,描述了汽車制造產(chǎn)業(yè)鏈上企業(yè)以不同方式的分類,以及企業(yè)之間供貨鏈的確定信息。在該空間中,粗糙數(shù)據(jù)推理的推演刻畫了企業(yè)之間潛在供貨渠道的粗糙數(shù)據(jù)聯(lián)系,提供了智能處理和自動管理的參閱信息,使粗糙數(shù)據(jù)推理的理論方法在實際中得到了的應用。討論了特殊的粗糙推理空間—樹型推理空間中的粗糙數(shù)據(jù)推理,展示了以樹作為推理關系的特點。在樹型推理空間中,利用樹包含的層次信息,證明了以樹作為推理關系的重要結(jié)論:粗糙數(shù)據(jù)推理的推演依賴于數(shù)據(jù)位于的層次。由此通過對樹型推理空間的細化,展示了細化粗糙數(shù)據(jù)推理更趨于精確信息的推理特性。同時細化粗糙數(shù)據(jù)推理的結(jié)論可用于汽車制造產(chǎn)業(yè)鏈上供貨依賴關系的分析,使理論方法進一步得到了應用。在粗糙推理推理空間中給出了粗糙路徑的概念,證明了粗糙路徑與粗糙數(shù)據(jù)推理之間的相互對應聯(lián)系,從而使粗糙路徑用于了粗糙數(shù)據(jù)推理內(nèi)涵精度的描述,由此區(qū)分了相同形式粗糙數(shù)據(jù)推理的相異內(nèi)涵,形成了對粗糙數(shù)據(jù)聯(lián)系松散或緊密程度的辨別方法,對于實際應用具有指導性的作用。通過結(jié)構(gòu)化的粒化樹構(gòu)建,并利用�；瘶渲械膶哟涡畔�,給出了數(shù)據(jù)關聯(lián)的定義,產(chǎn)生了粗糙數(shù)據(jù)推理的關聯(lián)推理方法。該方法以關聯(lián)數(shù)據(jù)作為橋梁,結(jié)合數(shù)據(jù)的等同、等同的更接近、數(shù)據(jù)關聯(lián)的形式、關聯(lián)情況的數(shù)值表示、關聯(lián)程度的極大性處理等,使兩數(shù)據(jù)類中的數(shù)據(jù)建立起了關聯(lián)關系,并以上近似的特定運算作為數(shù)據(jù)關聯(lián)判定的充要條件。該方法的特點體現(xiàn)了對�；瘶渲辛５膶哟魏土６茸兓膽�,以及對數(shù)據(jù)關聯(lián)和關聯(lián)程度數(shù)值表示的處理。同時討論與實際問題密切相關,基于粒化樹的數(shù)據(jù)關聯(lián)方法用于了具體問題的描述,實現(xiàn)了理論聯(lián)系于實際的研究預期。上述工作以粗糙數(shù)據(jù)推理作為研究的主體,以數(shù)據(jù)關聯(lián)推理作為研究的部分。探究步驟循序漸進,研究細節(jié)追求清晰、問題分析逐步推進、整體討論圍繞主題。這些工作包含了課題研究的自身方法,體現(xiàn)了對粗糙數(shù)據(jù)推理課題與數(shù)據(jù)關聯(lián)現(xiàn)象的理解與認識,形成了程序設計的算法基礎。同時針對實際問題的模型刻畫和實際數(shù)據(jù)聯(lián)系的粗糙數(shù)據(jù)推理描述,展示了理論方法源于實際,實際應用基于理論的研究目的。
[Abstract]:Study on information science involves all aspects of data processing, the related work to promote the direction of production, the results appear to promote the development of the discipline. As the direction of information science research or research data classification, data reduction, data warehousing, data filtering, data mining, data deduction for data processing program show the broad and active research field, but also contains the research concept of combining theory and application. Although different jobs have different emphases, but often involves the research level in common. Data is concerned, is not clear, uncertain, like the presence or potential data between the data associated with these directions are related at the same time, also appeared frequently in practice, which leads to the concept of rough data link. Thinking about this contributed to the rough data reasoning topic, less involved in the study indicates The significance and the frontier, and the theory research will provide the basis algorithm and program design based on rough data reasoning. This thesis focuses on the topic, complete the work focused on the following aspects: to rely on rough set approximation space was expanded on the structure, the reasoning relation, produced rough data reasoning to the implementation of the environment space. Relying on the rough reasoning described for connection to the data in the rough, rough reasoning space, approximate information fusion by equivalence relation and inference relation, leads to a rough number according to the definition of the reasoning, reasoning on data, the research topic. The research of rough data the reasoning, obtained the relevant conclusions, showing the nature of rough data reasoning including rough data reasoning keep determine characteristics of data relationship, rough data and reasoning On the approximate approximation characteristics is closely related to information, rough data reasoning has the characteristics of approximate description of function, characteristics of rough data reasoning and path are equivalent, rough data reasoning corresponding to different equivalence relation properties. Construct the rough reasoning of spatial problems, describes the automobile manufacturing industry chain enterprises to classification in different ways. And between the enterprise supply information to determine the chain. In the space, rough data of deduction depicts contact rough data between enterprise potential supply channels, providing intelligent processing and automatic management of the information, the application of theory and method of rough data reasoning has been discussed in practice. The special space rough reasoning tree type inference in space rough data reasoning, show the tree as inference tree inference relations. In space, the tree contains level The information proved to the tree as an important conclusion: the rough data reasoning of deduction depends on the data in the hierarchy. Thus through the refinement of the tree inference space, showing the characteristics of rough reasoning refinement data reasoning more accurate information. At the same time according to the number of refine the rough reasoning conclusion can be used for the analysis of automobile manufacturing industry chain supply dependency, the theory and method of further application. In the rough reasoning space gives the concept of rough path, proving the corresponding relation between the rough path and rough data reasoning, so that the rough path for the rough data reasoning connotation is described, which distinguishes the different connotation of the same form of rough data the reasoning, formed a discrimination method of rough data or loosely connected closely, is of great significance for practical application through the node. The grain tree construction, and using level of information granulation in the tree, gives the definition of data association, the association reasoning method of rough data reasoning. This method with associated data as a bridge to combine data equivalent, equivalent closer, data association, said the numerical Association. The correlation degree of maximal processing, so that the two data type of data to establish the relationship, necessary and sufficient conditions for a specific operation and above as approximate data association judgment. The characteristic of this method reflects the application of grain in grain and grain tree level changes, and the processing of numerical data association and said the association degree. At the same time discuss closely related problems, data association method for granulation tree based on specific description of the problem, the research realizes the connection of theory to actual expectations. The above work based on rough data reasoning As the research subject to data association reasoning as the research part. On a step by step, study the details of the pursuit of clear, problem analysis step by step, the overall discussion around the theme. The work includes research of its method, reflects the understanding and awareness of the rough data reasoning and data association problem phenomenon, forming algorithm based program design. Describe the rough data model to describe the relation reasoning according to practical problems and actual data, showing the theory stems from the practical application, the purpose of the study is based on the theory.

【學位授予單位】：北京交通大學
【學位級別】：博士
【學位授予年份】：2017
【分類號】：TP18
，

本文編號：1462218