Use of semi-automatic pre-treatments for term lemmatization and construction of contingency table.
Construction of a remormalized textual network from the text-document contingency table.
Comparison of Ncut and modularity clustering.
Construction of meaningful, well-defined terms communities.