Retrieval Technology For Requirements Document To Component

Software reuse technology is an effective way to solve the software crisis,and component reuse is the core of software reuse technology.Component retrieval technology is the core of component reuse.In conventional component retrieval technology,retrieval modes based on keywords,faceted description,and ontology are widely used.But there are also some problems in each retrieval mode.This article presents the improvement on the basis of the requirements document segmentation,analysis and retrieval methods,and achieves from requirements to component retrieval,which effectively improving the initial investment and insuring the recall and precision within a valid range.The main contents are as follows:1.This paper proposes a method of thesaurus classification,improving the accuracy of the most commonly used Chinese word maximum reverse approach.Then this article assigns weights to the terms and takes advantage of hierarchical entry mechanism to ensure that high-priority items can be accurately separated.2.To solve the problem of low accuracy of the matching degree between tree structure and net structure,this subject puts forward the ontology hits algorithm.Through the ontology hits algorithm,we improve the accuracy of matching degree between requirements tree and component relation net.3.This paper puts forward the description method of faceted-ontology component,which solve the problem of that single description method is easy to cause error and retrieval failure.The description method of faceted-ontology component effectively reduces the initial investment and insures the recall and precision within a valid range.4.This paper proposes the method of matching the mode based on document and the mode based on component to improve the performance,which make its performance such as the initial investment,recall and precision better than other frequently-used methods.This paper also experiments on a number of requirements documents in different systems to verify feasibility and performance indicators of this method.And experimental result shows that this method is feasible and effective.Compared with other current component retrieval methods,the performance of this method has improved to some extent,and its initial investment costs have reduced.