%0 Journal Article
%T Microblogging for Micro Sentiment Analysis and Opinion Mining Le microblogging pour la micro analyse des sentiments et des opinons
%A Alexander Pak
%A Patrick Paroubek
%J Traitement Automatique des Langues
%D 2011
%I Association pour le Traitement Automatique des Langues (ATALA)
%X Microblogging is a recent trend in today's Internet. Users express their opinions using microblogging platforms such as Twitter. In our research, we use Twitter as a multilingual data source to collect a corpus of sentiment labeled texts. Using the lexicon extracted from our corpus, we build a sentiment classifier which we apply to three kinds of tasks: classification of sentiments in short texts, disambiguation of sentiment ambiguous adjectives, and construction of affective lexicons in different languages. We call them amicro sentiment analysisa tasks as they operate on small texts or spans of texts. Experimental evaluations using hand-annotated dataset, participation in the SemEval 2010 evaluation campaign and correlation with ANEW affective lexicon prove that our method performs well, even if we do not use language specific tools and human-built resources.
%K sentiment analysis
%K microblogging
%K Twitter
%K SemEval
%K ANEW
%K multilingual approach
%U http://www.atala.org/Le-microblogage-pour-la