Comparing Dictionaries for the Automatic Generation of Hypertextual Links: A Case Study

There is a great need for tools that can build hypertexts from "flat" texts in an automatic mode, assigning links. This paper addresses the problem of the automatic generation of similarity links between texts that are relatively homogeneous in form and content, such as the cards of an art catalogue. The experimentation it describes has compared the results obtained using weighted and unweighted supervised dictionaries with those produced using weighted and unweighted automatic dictionaries.