基于构件可信性的构件刻面描述与检索研究

发布时间：2018-03-17 23:21

本文选题：构件可信性　切入点：可信性评价模型　出处：《昆明理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着计算机技术发展的不断深入,计算机软件已经成为人们生活必不可少的部分。由于软件的特殊性,无法像工厂一样流水线生产,提升软件的生产效率与质量是相当重要的,基于构件的软件开发随之而诞生。通过基于构件的软件开发技术,人们认识到通过选取已有的构件去组装系统,不仅能够提高工作效率,而且还能减少开发费用。但盲目地选取构件去构建软件系统,必然会产生构件可信性问题,规模越大的软件系统,越让人们缺乏信任,可信性问题日益突出,所以选取可信的构件是构建软件系统的基础。如今,构件的可信性渐渐地成为构件管理的研究重点,构件技术发展的研究方向。虽然国内外的研究人员都对构件的制作、管理、复用等阶段进行了许多实验性的研究,但对构件管理和复用阶段的可信性研究还很少,由于国内外文献对构件可信性研究没有进行统一模型的定制,本文综合权威文献,详细地分析研究了构件可信性的七个特性及其子特性,给出了子特性的度量模型。建立了构件可信性评价模型,运用模糊综合评价方法进行构件可信性的评价,利用模糊理论可以解决构件评估过程中的不确定性及模糊性。并在构件现有的刻面分类描述的基础上,增加了一个新的刻面——可信性刻面,有利于构件的全面描述和准确检索。在传统的构件检索中,是将构件库中的每一个构件转化为对应的构件描述树,同时将用户输入的检索条件转化为用户所要求的构件查询树,最后将两者对比进行匹配。本文提出了树匹配与刻面权重相结合的构件检索方法,首先取出构件查询树与描述树的可信性数值向量;然后运用Cosine相似度计算方法,从构件库中筛选一部分相似度较高的构件;接着再通过筛选出来的构件集合,对每一个构件设置刻面编码树;最后以字符串匹配来进行构件的二次检索,避免了大量的检索工作,提高了构件检索的效率。并且在匹配度计算时,把构件刻面权重考虑进来,以提高构件检索的科学性和有效性。根据本文所运用的构件刻面描述与检索方法,实现了一个构件库系统原型,分析了不同构件查询下的查全率、查准率及检索效率,验证本文所运用的树匹配与刻面权重相结合的刻面检索方法的正确性与可行性。
[Abstract]:With the development of computer technology, computer software has become an indispensable part of people's life. Because of the particularity of software, pipeline cannot produce it like the factory, so it is very important to improve the efficiency and quality of software production. Component-based software development is born. Through component-based software development technology, people realize that by selecting existing components to assemble the system, not only can work efficiency be improved. But blindly selecting components to build a software system will inevitably create a component credibility problem. The larger the software system, the more people lack of trust, and the credibility problem becomes increasingly prominent. Therefore, selecting credible components is the basis of building software system. Nowadays, the credibility of components is gradually becoming the research focus of component management and the research direction of component technology development. Although researchers at home and abroad make and manage components, Many experimental studies have been carried out in the reuse stage, but there are few researches on the credibility of component management and reuse stage. Because there is no unified model customization of component credibility research in domestic and foreign literature, this paper synthesizes authoritative literature. In this paper, the seven characteristics of component credibility and its sub-characteristics are analyzed in detail, and the measurement model of sub-characteristics is given. The evaluation model of component credibility is established, and the fuzzy comprehensive evaluation method is used to evaluate the credibility of components. The uncertainty and fuzziness in the process of component evaluation can be solved by using fuzzy theory, and a new face-credibility facet is added on the basis of the existing facet classification description of the component. In the traditional component retrieval, every component in the component library is transformed into the corresponding component description tree, and the retrieval condition input by the user is transformed into the component query tree required by the user. Finally, this paper presents a component retrieval method combining tree matching and facet weight. Firstly, the credibility value vector of component query tree and description tree is extracted, and then the Cosine similarity calculation method is used. Filter some components with high similarity from the component library; then set a facet coding tree for each component through the filtered component set; finally use string matching to carry out the secondary retrieval of the component. It avoids a lot of retrieval work and improves the efficiency of component retrieval. In the calculation of matching degree, the component facet weight is taken into account. In order to improve the scientific and effective of component retrieval, a prototype of component library system is implemented according to the method of component facet description and retrieval, and the recall rate, precision rate and retrieval efficiency of different component query are analyzed. The correctness and feasibility of the facet retrieval method combined with tree matching and facet weight are verified.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.52

【参考文献】