Web-based Multimedia Data Mining Research

Data Mining and Knowledge Discovery provides us a new way to resolve the problem that we can't make the best of increasing huge data. Multimedia Data Mining is the technique to be used to discover the implicit, effective, valuable and intelligible pattern from a large amount of multimedia data by analysising the feature of seeing and hearing, and then to discover knowledge and obtain the tendency and association among the events. And it can also provide us the ability of decision-supporting to resolve the problem.The target of the paper is not confined to designing the algorithm, but studying deeply how to improve the function of DMS. And a Multimedia-Data-Mining model is put forward, which is based on all kinds of data in the web.In accordance with the international and domestic development and tendency of MDM, the paper discusses the basic theories of Data Mining and Knowledge Discovery, then analysises the feature of information retrievaling. After the theory and technique, a serial of questions on Multimedia Data Mining are explained in detail. At last, a Multimedia-Data-Mining model is present, and the function of each module is introduced effectively.This model can retrieval information on the web page automatically through concept retrieval of multimedia, then save the data into the feature-base. Receiving the ask for extracting, the model can transmit the result to user after carrying on retrievaling by vector space model.The paper has done the following work mainly:1 .Discussing the basic thought and technology of Data Mining and Web Data Mining.Data Mining is a procedure that extracts the potential , valuable knowledge (model or the rule ) from a large amount of data. It is the core that researches various kinds of methods and technology of knowledge discovery mainly. Web Mining is the course of collecting information or knowledge from the Web. It applies common thought and method of data mining to Web.2. Studying characteristic and method of multimedia data miningBecause of the feature of Semi-structured or none-structure , the way to mine the multimedia data in web doesn't equal to common data mining. The paper discusses how to transform the Semi-structured or none-structure data to well-structure.3.To the picture data, this paper designs an integrate comparatively multimedia feature base.Feature extracting is the key component of CBIR system, the quality of Feature extracting is important for CBIR system, it concerns the accuracy of the system. A Multimedia Data Mining model based on web is put forward, and the function of each module is also introduced effectively, which founds a good condition for next concrete development and research.