Abstract : The automatic clustering of text segments into thematically homogeneous groups is a difficult problem. In this paper, we study the performance of a simple probabilistic model, the "monomaniac" model. We first describe the model and the related EM-based estimation procedures; an application of this model to a English corpus of texts imported from the CKM ("Customer Knowledge Management") literature is then presented.
C. Marty. Amplitude reconstruction in charged particles scattering. International Workshop On Advanced Methods In The Evaluation Of Nuclear Scattering Data, Jun 1985, Berlin, Germany. pp.223-228. ⟨in2p3-00005962⟩