Research And Implementation On Technology Of Component Description And Retrieval Based On Functional Semantics

Software reuse is one of hotspots in the field of computer software research and development. It can increase software quality and developer productivity, decrease the cost of development and maintenance. Component-based software development(CBSD) is claimed to be an effective form of software reuse, while the effective way of component description and retrieval is a key technology. The traditional component description and retrieval approaches have the problem of lacking the semantic description ability, which affect the recall ratio and precision ratio. Therefore they can not achieve the aim of software reuse.Ontology technology is able to solve the problem of lacking semantic in existing component description and retrieval methods. This dissertation analyses the component description and retrieval based on facet schema, and combines it with ontology technology. An approach of building component ontology with function facet is proposed. The improvement of component retrieval based on this ontology is proposed, too.Main contributions in the thesis are:Firstly, study the current approaches of component description and retrieval ,analysis different facet characteristics and their contribution to description and retrieval, research on the term characteristic under the function facet and the semantic dependency relationships which were classified into functional dependence, data dependence, control dependence and communication dependence as general functional semantic relationships. Non-functional relationships of component were also studied and classified into similarity relation and hierarchical relation as general component semantic relationships. The associations among these relations were extracted into rules as the basis of semantic inference. Secondly, an approach is proposed to build component ontology with function facet. The approach maps the the terms under function facet to the concepts in the ontology model and describes the relations between concepts with functional semantics relationships, such as functional dependent, data dependent etc. A formal definition of this ontology model is also given. Credit component library is taken as an example to explain this approach.Thirdly, on the basis of component ontology, approaches to calculating the semantic similarity between two concepts and components are proposed. On the basis of semantic relation among concepts and among components, approaches to calculating the semantic relativity are also proposed.Finally, with the approaches mentioned above, improvement is made on the Credit component library based on facet classification, providing semantic support for component description and retrieval.