Automatic metadata generation based on neural network

Metadata technologies can allow the proper search and processing of Web pages, which has been used in many fields, such as e-commerce, library science, web search, and so on. In the approach, a novel framework is proposed to automatically generate Dublin Core metadata about web pages, what is finally expressed in XML. The framework is located on a web server, where can provide detailed information about web pages to generate the metadata. Moreover, the framework utilizes a combination of a well-trained neural network and traditional statistical methods to filter proper value for the metadata. Experimental results show that the proposed framework is preferable to the traditional statistical method.