Research on Information Querying, Browsing, and Filtering Techniques for Electronic Commerce

As the internet becomes more popular, it is easy to provide and acquire data through the internet. On the world-wide-web, a huge amount of data is crowded, along with emerging users. Thus, more diverse (variant) and high quality information services are needed. In this project, we will focus on research issues for providing information services under electronic commerce infrastructure, including information querying, information browsing, and information filtering, which are described as follows. 

1.   We will construct a multimedia directory system, which is equipped with the ability of content-based retrieval and powered by personalized query suggestions along with users¡¦ operations. In addition to analyzing the multimedia content of product information, and extracting multimedia features to construct indexes and develop query processing algorithms, we will construct distributed multi-users system. A web-based and user-friendly interface will be also provided for customers to browse product information in the directory and to pose content-based queries. By analyzing users¡¦ access log to the system, such as query history, click-through, and so forth, we will generate user profiles. According to user profiles, the structure of directory will be adjusted dynamically to provide a suggested browsing in a personalized way. Moreover, we group the users and produce buying recommendations. When evaluating queries of a user, the query predicates will be rewritten referring to his/her profile. The query results will be ranked accordingly such that those most-wanted results will be returned first. 

2.   We will analyze web-pages visiting history and transactions of users to provide a flow-based browsing. The flows which have rich semantics will be derived. Customers can query the flows. Following the flows, customers will quickly find information of wanted products such that business transactions can be increased. The flows are task-oriented and have large supports. Beyond existing search engines on the internet, the flows will better satisfy the requirements of customers. Moreover, by considering the ordering of the web-pages in the flows, we will improve the flows to closely meet browsing behaviors. By deriving the flows, we will provide users the information browsing services, including creating a database for the flows, providing query facilities of the flows, and then adjusting mechanism. 

3.   Form the browsing behaviors, transaction information, and customer interests, the non-interested product information can be filtered out. Since every customer is an individual and has his/her own needs and preferences, the information filter should be performed in the fully personalized way, making use of the filter profile, customer feedback, topic network, and so forth. We will implement a system to provide automatic information filtering services. The system will learn customer interests, adjust filters to help customers to gather most related product information. Besides, by comparisons of topic networks, customers will be clustered into some groups. The customers in the same group will share the recommendation from the system.