In this paper we present a description and evaluation of a word sense disambiguation mechanism that is an extension of the work by Ide and Veronis. The mechanism's underlying approach is a numerical constraint satisfaction algorithm applied to the network built using an on-line dictionary. In this work we aim to duplicate the original results and extend the method to large-scale applications. To determine the scalability of the approach, we evaluate the mechanism on 20 ambiguous words taken in their contexts of different size from the Brown corpus. We offer a discussion of encountered difficulties and describe several possible directions that can be investigated to improve the mechanism's performance and scalability.