Web-based information access: multilingual automatic authoring

The need for managing similar documents in different languages is increasing with the growing amount of electronic information that is available in documents of the same type (e.g. news streams). This paper proposes a viable approach to information access, emphasising the hypertextual paradigm in a multilingual framework. This task of processing/structuring text so that cross-lingual hypertext links are generated is called multilingual authoring (MA). Methods from natural language processing, especially information extraction, to both monolingual authoring and MA are described, and a general architecture for MA is defined. The effectiveness of the proposed approach is discussed using the description of the NAMIC (News Agencies Multilingual Information Categorisation) prototype system.