Component Retrieval Based On Ontology

Component-based software development (CBSD) is an appealing approach for industry-scale software construction. CBSD promises not only improving software productivity, reducing the time-to-market and catching business opportunities, but also differentiate software manufacturer from others in terms of customer satisfactory and software quality, as well as making software projects successful. Component technology has become the cornerstone for software industry makes progress. It has become a worldwide trend to adopt CBSD in software application development process.The description and retrieval of components is one of the key technologies to actualize the potentiality of component technology for software reuse and construct high quality software products. And it is become more and more important with the increase of the number of components. But current methods for the description and retrieval of components still leave much to hope and become one bottleneck in the development of the component technology.Based on ontology technology, the description and retrieval of components can obtain more satisfied effect in experience and people place much hope on this. But the study of this is only beginning and there is much work to do and the paper does some attributes for this: a component description model is given based on combining ontology domain and feature domain model; a component retrieval algorithm and an algorithm to retrieve related component based on the description model; a method to assist the user to refine his retrieval requirement during the phased component retrieval; a component retrieval model based on Dominance-based Rough Sets Approach.First, the paper gives a component description model based on combining domain ontology and feature domain model. The component description model contains function attributes, non-function attributes and QoS attributes. Then how to index and organize components based on the component description model and the ontology feature domain model is given. Based on those, component retrieval algorithm and an algorithm for retrieving related components to the components retrieved are introduced respectly. Experiment demonstrates that the work give good component retrieval result.Second, a method to assist the user to perfect their retrieval requirement is intro- duced. First, system assists the user to complement verbs represented function action or nons represented business object and adjust the function action comtained in his retrieval requirement. Then through mining the relation between the item values of the retrieval requirement, the system assists the user to complement missing item of the retrieval requirement. Take the corresponding item values of the candidate components found thrugh component index as candidate valus, and return to the user to deal missing items or items with wrong value of the retrieval requirement. Experiment and analysis demonstrate the retrieval requirement refining assistence method can improve the retrieval result.Third, to deal with some problems in component retrieval, a new component retrieval method is given based on Dominance-Based Rough Set Approach (DRSA). See the items in past component retrieval requirement as condition attributes (or Criterion), and the feedback component grades as decision attributes. The system learns rule knowledge based on DRSA. And in later component retrievals, the system obtains retrieval result based mainly on rule with the auxilirary similarity counting. Experiment and analysis demonstrate the the method can obtain better retrieval result.