Design And Implementation Of Domain-specific Business Information Retrieval System In Electronic Commerce

With the prosperity of computer networking technology and internet, Electronic Commerce (EC) becomes more and more popular in many enterprises around the word, and it has become the first choice for competitive strategy of enterprises in business trade. With the intensive EC applications in enterprises, enterprises need more and more business information about raw materials, suppliers, customers and products to support the decision-making on purchase. The enterprises should go beyond the relatively narrow business environment in the past, to collect and use business information effectively. So, business information has been necessary for enterprises to succeed in competitions.Taking the project of "The first period of logistics system optimization in Shanghai Bus group" as research background, based on in-depth analysis of business information problem in EC and reading lots of related literature, this paper proposed a business information retrieval model based on extended belief network firstly, then focused on designing and implementing a domain-specific business information retrieval system in EC environment. The major contents are as follows:(1) With regard to the problem of low recall and precision in business information retrieval, an extended belief network based business information retrieval model has been proposed. Considering the features of business information and effect of relationship among index words to retrieval performance, this model extends the belief network with four layers of nodes, and applies similarity thesaurus to expand the user query and describe the relationship among index words. The similarities between documents and query can be evaluated by calculating the conditional probabilities among the nodes. The experiment results show that the proposed model has a good performance for collecting business information.(2) Based on analysis of the common methods of business information retrieval in EC environment, a software system has been designed, which can automatically collect business information in Internet and extract business information required by enterprise from database directly. This system adopts meta-search engine to expand search scale, and applies information retrieval, web mining and agent technology to analyze and filter the business information, in order to improve the search quality of business information. (3) In order to verify the design effect, a business information retrieval system has been developed with respect to the domain of auto parts and integrated with three famous GSE: Google, Baidu and Sogou. The operation effect verifies the feasibility and effectiveness of the system design.