The Research On Component Library System's Semantic Expansion And Ontology Reasoning

With the rapid development of the software industry and the widespread use of software, software development has been faced with a growing number of challenges. In this context, reusable software component library as an important basic facilities of software reuse has increasing arouse the industrial circles'and the academia's attention. As the core of component library is the retrieve mechanism, therefore, to find an efficient retrieve algorithm for the component library management system and improve the retrieve's precision and recall ratio are particularly important.In recent years, with the research and development of ontology technology, the scholars home and abroad focus on rising the level of inquiry from keyword-based matching to the knowledge-based, semantic-level matching and giving the knowledge processing capacity and the understanding ability for component library management system to achieve the intelligent inquiry.Originally, ontology was a concept of philosophy, which was used to describe the nature of things by philosophers. Since 1990's, the concept of ontology was introduced in artificial intelligence, knowledge engineering and the field of information retrieval, so that the connotation of the concept of ontology also changed. A clear definition of ontology has undergone a process; finally, a definition of ontology given by Gruber has been acknowledged by the industry which describes"The ontology is an explicit specification of a conceptualization". Generally speaking, the research on ontology is about the representation of knowledge concept and the knowledge organization systems, namely, is about the vocabulary terms on a field and the norms and descriptions of the relationship between terms. At present, the research and application on ontology are showing the trend of accelerating development. The basic theories become more and more complete, and have been widely used in application areas, especially in knowledge-based information retrieval.Based on the research of ontology and its related technologies, a component library management system framework based on semantic expansion and ontology reasoning is raised, in order to maximize to meet the users'demand of the inquiry components and improve the precision ratio and recall ratio. The tasks can be specific as follows:First: using the OWL ontology language to establish a reasonable component model.OWL (Web Ontology Language), which is formally launched in February 2004 by W3C, is a specialized language used to describe the ontology. OWL can be used to store the concepts and relations; it has a strong ability to express semantics and can realize the decidable reasoning of description logistic.This paper uses the OWL ontology description language to describe the domain of component, proposes a component description model based on faceted classification. The model divides the component description information into the basic information, classified information, interface information and the relationship information. The establishment of component library ontology semantic networks lays the foundation for research of component ontology reasoning.Second: using RDF ontology description language to describe the information of component instances.RDF (Resource Description Framework) is a resource description framework language, which can be used to describe the meta-data and the relationship between meta-data. Meta-data is data about data, is the data to describe the network resources. This paper uses the RDF description language to describe the information of component instances and store the information into the file system.Third: expanding the content of user inquiries on the semanticIn order to reduce the issues such as semantic loss and distortion on the conversion process from the user's inquiry intention to the inquiry language, this paper has made a corresponding introduction on the inquiry expansion technique, mainly about the advantage of expanding the user's inquiry information contrasting with the conventional inquiry dispose method, and the semantic-based inquiry expansion method, and the definition of similarity, and the degree of similarity and the degree of correlation between concepts.In this paper, "WordNet" is the basis of measuring the degree of similarity. It is an English dictionary based on mental language rules. It is considered as the most important and quality research resource which can be obtained in the region such as computational semantics and text analyses. It uses semantic Web as the basic representation form of its vocabulary ontology. In this paper, WordNet is used to measure semantic distance between the concepts, and finds out the degree of correlation between them. Studies have shown that distance is an important lexical relation between words and there has a close relationship between the lexical distance and lexical similarity. The longer the distance of two words, the smaller the similarity is; on the other hand, the shorter the distance of the two words, the greater the similarity is. We can build up a simple corresponding relationship between them.On the concrete realization, we measure the matching degree between the input content from the user interface and the describing information of the component which is stored in the component library. And then, a threshold will be set and all the component describing information whose score is larger than the threshold is considered to be correlation of the user inquiry content and be entered into the result set. By doing this, we achieve the semantic expansion on user's inquiry content, and improve the recall ratio. At last, the result set will be returned to the user interface in order to further clarify the inquiry intent to prevent the semantic expansion deviate the user's inquiry subject.When we measure the degree of similarity between the inquiry content and the component describing information, mainly, we refer the method on measure similarity between sentences on the paper Measuring Similarity between Sentences, which wrote by Thanh Ngoc Dao and Troy Simpson. It mainly related to vocabulary semantic disambiguation, a similarity measuring method based on the length of the path for measuring the semantic degree of similarity between two synsets on WordNet, and the method to measure the similarity between sentences.Fourth: using the Jena ontology reasoning engine to reason the component ontology.Jena is an open resource from the semantic web research projects by HP Labs. It provides a programmable environment for ontology description language: RDF, RDFS and OWL, and provides the analytic, endurance memory, reasoning and inquiry interface for the function call and process.In recent years, the work on ontology reason has aroused much attention. The research on ontology reasoning becomes more and more mature. Through reasoning, people can gain knowledge from the ontology. Practice shows that some important knowledge is often hidden in the ontology. Therefore, it has an important significance to research on reason the hidden knowledge from the ontology, especially from the ontology based on OWL description.In this paper, through the research on ontology reasoning and relevant technical studies, we realize an ontology reasoning framework based on Jena. We use the reasoning sub-system of the Jena to reason the component library ontology based rule reasoning machine (OWL Reasoner). It can be divided into two main approaches, one is, when a user does not provide rules for reasoning, the system uses the Jena's own general rules reasoning machine for reasoning; the other one is, when the user provides the rule according to their needs, the system will reason by the user-defined rules.Fifth: after summarize the whole paper,we raise some problem which need further research.Finally, through applying the ontology and its related technology, we develop a component library management prototype system based on semantic expansion and ontology reasoning framework. The experiment shows that: this system can regulate and expand the user's inquiry request intelligently, and can excavate the implicationrelationship between the components, and improve the efficiency of component retrieval, and improve the retrieval's precision radio and recall radio, and provide an effective support for component library's intelligent retrieval.