Ameritech needed to migrate a large set of legacy M&P and other data to a new content management system, which was to support a new web-centric call center help application.

The legacy content consisted of 10,000 pages of unstructured mainframe textual data. It also contained ad hoc "character-oriented" formatting of complex elements, such as tables. No coherent taxonomy existed for the legacy data. The nature of the call center application required an intuitive search engine to quickly locate relevant documents.

Pattern recognition techniques were applied to automatically transform legacy documents to XML and HTML. Intelligent Systems applied clustering techniques to automatically construct a hierarchy. Finally, a custom search engine that closely modeled user expectations was designed.