You are here
Text Mining, Dreams and Elections
In this talk, we will review some of the recent applied text mining work at Dalhousie. We will argue the need for a text representation that would be more linguistically informed than the standard vector model. We will present one such proposal, in which a co-occurrence model takes into account the distribution of words throughout the corpus. We will then show how this representation is successfully applied in the task of categorizing dream descriptions by their emotional valuation (joint work with J. De Koninck and A. Razavi, Ottawa). We will round up the talk with our experience with some of the other text mining techniques used in the analysis of the twitter traffic in the 2012 presidential elections in France and in the US (joint work with LIRMM, France).
About the Speaker:
Stan Matwin is a Professor and Canada Research Chair at Dalhousie University, and a Distinguished Professor at the University of Ottawa (on leave). Fellow of ECCAI and CAIAC and an Ontario Champion of Innovation. Internationally recognized for his work in text mining and in applications of Machine Learning, member of Editorial Boards of the leading journals in Machine Learning and Data Mining. Stan Matwin is one of the founders of Distil Interactive Inc. and Devera Logic Inc., and has significant experience and interest in innovation and technology transfer.