Factor PD-Co-clustering in Official Statistics

In this paper we propose an extension of Factor PD-clustering for simultaneous classification of rows and columns of frequency matrices extracted from largetextual datasets. The aim is to extract information from documents usually producedand that are not used because of their special nature. The work is carried out withinthe European project BLUE-ETS, which aims to provide tools for the constructionof robust and high quality official statistics for businesses.