The description of Web resources is one of the problems related to the development of the Semantic Web. The major problem in metadata generation is due to lack of experts and tremendous amount of web resources. It is expected that an automated system would encourage creation of metadata for Web resources. This work is focused on suggesting four classes of descriptive elements – bibliographic, semantic, keywords, structure – in designing an automated metadata generation system. In addition, this study tries to assess how the level of knowledge or skill impacts the quality of the metadata generated and based on the results it suggests factors that automatic metadata generation tools should suggest to users.