Information Retrieval With Varying Granularity On Structured Web-course Document

With the fast development of web-based education, the construction of tech resources became the more important factor. In the construction of tech resources, whether carry through fast and exact searches and sharing and reuse or not, which became key problem of resources take maximum function. Retrieval of information is the base of information search, information sharing and reuse. Since there have been a lot of courseware on the web, how to retrieve useful information from the large amount of courseware for the using of information in the construction of other courseware or other applications became the key problem of these tasks. Because of the particularity of the network courseware, its main contents are made up of amounts of multimedia information. Traditional information retrieval methods are mostly based on the study of retrieval of text information, so to design the new algorithm to retrieve multimedia information is necessary. Traditional information methods always return the whole document or a full paragraph to the users as results. But in course of query, information sharing and information reusing, users usually require the result to be more exact. If the information retrieve method can provide multi-granularity choose of the information retrieve result, it can give user more efficient information, shorten the time of re-query the information from the results and provide more suitable results for different user or applications. This paper proposes a multi-granularity information retrieve method for web-based courseware. With the prevalence of remote education, the standardization problem has become the hotpot problem and attracted more and more attention over the world. Some related institutions are making serials of rules and standardizations. In our state's related standardization, the standard courseware model has been proposed and XML has been the stated description language for web-based courseware. In the recently future, all web-based courseware will follow the standardization, and the documents of courseware will be structured documents. In order to be compatible with the future standardization, multi-granularity information retrieval for web-based courseware become the multi-granularity information retrieval for structured document problem in fact. Based on the web-based courseware standard model, the paper proposes the retrieval model for the information retrieval. The retrieval model has changed the standard model at some degree. Based on the retrieval model, designs the multi-granularity retrieval procedure and related algorithm for web-based courseware. The algorithm is mainly about of two aspects: multi-granularity and multi-media retrieval. In the multi-media retrieval, a threshold is used, and in multi-granularity we make reference to graphic theory methods. For implement and evaluation of the multi-granularity information retrieval, an experiment system has been developed. Some experiments have been done to find out the result of changing main variables. The experiment shows the method can get more exact results and some suggestions are made for improvement in future research.