| Abstract [eng] |
Web document automatic tagging isn’t very widely used today. But because of significantly increasing number of documents, this area of natural language processing attracts a lot of attention. Automated web document tagging could help users to quickly found information. It also can connect texts that are written by different authors. In order to determine possibilities of web document tagging system, possible analysis algorithms are discussed. Programming models, common problems of natural language processing and required steps that need to be done to create automatic tagging system, a review of decisions, provided functionality is described. The paper provides implementation of the web documents automatic tagging system and a description of its essential features. Performed experiment provides information how different algorithms behave, what metrics they have. Experiment also tries to investigate how different text length and context affects precision and recall. |
