Enhancing a hypertext application using NLP techniques

Many multimedia applications use the concept of hypertext as a means of relating textual components with associated data in a vanety of media forms. The hypertext concept facilitates one-to-one matching between lexical items (words) in text with equivalences in other data sources. User interface software thus provides access to related data by means of keyword or phrase matching techniques which are essentially lexical in nature. The problem with this approach is that a user is generally presented with a vast amount of information across a wider domain than is really required and which is often difficult to interpret. In this paper, we descnbe four natural language processing (NLP) techniques to improve the quality of the information presented to the user in a hypertext application called CDWord™