Component Classification And Retrieval Based On Facet Classified Description

Software reuse is a feasible approach to improve the development efficiency and the quality of software, especially under the current condition of financial crisis. Component library, as the infrastructure of software reuse, become frontier and hot point and receives much concern from both academic circles. Moreover, the representation and retrieval of software component is the hotspot of the component library research.With the deepening practice of software reuse, the reusable component not only limit to the source code but also extend to other product during software development, such as requirement analysis, design, test cases, etc. There are already a few classified description and retrieval methods for effective management of the reusable component, but single retrieval mechanism has inherent limitation. On the foundation of facet classification, this thesis presents an interactive component retrieval model based on feedback mechanism.Firstly, representation method and similarity calculation formula are presented based on the characteristic of facet classification, so the component retrieval problem is turned into the most similar component searching problem. Then, components are classified by cluster analysis algorithm and those with similar function are merged into one category. On this basis, with the idea of relevance feedback and application of Ricchio query vector transfer algorithm, better query vector toward the ideal query point is found by repeated interaction with the user.Finally, a prototype system is implemented to carry out this retrieval method. The feasibility and validity of the method is further proofed by the comparison of experimental data. This study on the component retrieval has a reference value and practical significance.