Research On The Theory And Method Of E-Catalog Ontology Self-learning Oriented On Customer

With the rapid development of the semantic Web technology and e-business applications, users have to spend a lot of time in various heterogeneous, obscure ocean in searching the catalog information required, and there are bottlenecks in semantic interoperability and integration when enterprises exchange the information. Normal, explicit specification of a shared conceptualization can be a good solution to this problem. In addition to meet the user's query, e-Catalog ontology also facilitates computer information exchange, semantic search and information identification. But establishing the ontology necessitates a great deal of expert assistance; manually setting it up would entail a lot of time, not to mention that there are only a handful of experts available. For this reason, using automatic technology to construct the e-Catalog ontology is a subject worth pursuing.Based on the above issues, this paper draws on the existing research results of ontology self-learning and e-Catalog international classification standards, ontology and semantic Web theory, and use natural language processing methods and techniques, in order to research customer-oriented e-Catalog ontology modeling and self-learning methods. The main work includes:(1) Constructing e-Catalog ontology meta-model. E-catalog ontology meta-model provides a unified domain framework for the ontology self-learning. This paper semantically extended the international classification standards combined with customer demands, and designed e-Catalog ontology meta-model with four levels, and described it with standardized language.(2) Automatic construction e-Catalog ontology based on meta-model:●Catalog Website level-based e-Catalog ontology concept acquisition. On the analysis of e-Commerce Web site structure maps and Web page block, this paper designed catalog ontology concept acquisition algorithm, and further use international classification standards to standardize the concepts.●Semantic and association rules-based ontological relationship learn. This paper proposed the method of acquainting e-Catalog ontology taxonomic relationship based on pattern matching. Verb is important in the non-taxonomic relationship, the paper also provided the method to mine the non-taxonomic relationship based on semantic association rules.●Pattern matching and on-line statistics-based ontological properties identification. This paper proposed using pattern matching and Web statistical analysis to identify the properties, and in particular, analyzed the property values and measure of data property.●E-Catalog individual extraction. Because e-Catalog ontological individuals lie in the leaf node of e-commerce site, this paper proposed an algorithm of automatic extraction ontological individuals.(3) E-Catalog ontology self-learning empirical analysis. This paper took Amazon.com as an example to show the process of e-Catalog ontology self-learning. Finally the paper concluded that the self-learning method is effective and valuable by the evaluation method.