Study Of Product Classification Ontology Building, Reasoning And Mapping

Product classifications can be used to define and describe the technical parameters of products. They provide a structure of the product classification code, integrate the process of product inquiries, product launches and product data collection, which make product information easier to understand and process. These classifications set a unified standard for products manufactured by different organizations and factories, making the real-time information of products systematic, consistent and shareable. In the age of big data, the commodity trading occurs all the time, which produces vast amounts of data associated with the product. To process the large amounts of data, there are two possible ideas:one is based on the technology of data mining. By data mining can help us identify the intrinsic links hiding in those non-standard, unstructured, semi-structured data. Another method is based on information organization, that means we should standardize the data in advance. Following this line of thought, product classifications emerged. However, the product classification has five inherent defects:the conflict between the linear and network structure, the limited ability to express semantic information, the limited specificity of concepts, the long update cycle and the too professional instructions for use.With the further acceleration of the process of economic globalization, all trading application systems have strong demand for information exchange and resource sharing. However, due to the situation that the current product coding systems used by countries in international trade are not heterogeneous, resulting in different product data structures varying, which leads to the phenomenon of "islands of information" and restricts the efficiency of information exchange. Currently, there have been a number of widely used international product classification code systems, such as GPC and eCl@ss.. The five inherent defects of product classifications have restricted the further use and spread of product classifications. But the research progress of the five issues is slow, many domestic and international scholars mainly study about how to construct product classification ontologies. In their studies, the construction process does not involve any complex semantic, which leads to the poor reusability of ontologies, At the same time, the inherent defects of product classification still have not been effectively solved. Therefore, in order to figure out the inherent five issues of product classification systems, this paper carries out the related research.Firstly, this paper introduces the formal concept analysis into the construction process of the product classification in order to improve the efficiency of ontology construction and rule out the subjective factors in the construction process to some extent. This paper explores and proposes a construction method for product ontology based on formal concept analysis and applies this method in the construction of ontologies based on GPC and eCl@ss. This step solves three defects mentioned above:the conflict between the linear and network structure, the limited ability to express semantic information, the limited specificity of concepts..Secondly this paper puts forward three standards to judge whether the product classification ontology is satisfied and then extends the existing product classification ontology query model. On this basis, this paper proposes semantic query model of product classification ontology and realizes the desired product classification ontology query mode by programming, extends the query from the previous Intra-class query to the classes with attributes and attribute with attribute and class with instance, which solves the problem of too professional instructions for use. Finally, this paper studies the semantic Interoperability between GPC and eCl@ss in the fourth chapter and proposes a complete product classification ontology mapping process based on semantic analysis. Then we select the appropriate mapping tools and do an empirical research of product classification ontology mapping. Next, we use query system of the fifth Chapter to evaluate the ontology mapping process, it greatly makes up for the shortcoming of long and slow update cycle. So far, the five issues of product classification that this paper proposes have been effectively solved. In addition to this,Tbis paper also accomplishes the construction, query and mapping of product classification ontology and realizes the reuse of the product classification ontology.The construction of product ontologies is the process of product knowledge representation. Therefore, the construction method should meet three requirements. First, the ontology must fully express the architecture of existing product classifications. Second, the ontology must have a strong ability of expressing semantic information. That is to say, the ontology can fully express the various semantic relations contained in the product classifications. Last, the reasoning should be judged, that means, the ontology can give a correct result of the query within a limited time.To deal with the defects of the production classification and the problems existing during the use of these classifications, this paper uses formal concept analysis to identify the inherent semantic relationship of the concepts, describes the relationship between formal semantics with a logical tool and then construct the product ontology, which realizes the transform from form matching to semantic matching. In addition to this, this paper also provides mapping tools for different classifications, which improves the user’s query efficiency.The product ontology constructed in this paper is not intended to replace the existing product classifications, but as a bridge between the user and the product classification. With help of our ontology, users’ queries can be mapped to the product classification by ontology mapping. Additionally, our method also optimizes the existing product classification systems and improves the inherent problems of the classifications. Last, by ontology mapping, we build bridges between different product classifications, which makes semantic interoperation available.Although this paper did a lot of basic researches related to product classification, further examinations are necessary for practical application. For example:Except the category of computer and communications we used, do our methods suitable for other categories? How is the quality of the mapped ontology? All these issues are worthy of further studies.