Abstract: Millions of Facebook and Twitter users send their comments all over the world about products and services, political and economical events, etc. (almost 3 billions each day) The principal problem of opinion mining such information is text parameterization, and in the paper we describe our experience in solution of this problem with Peruvian Facebook and Twitter. We use enriched vocabularies of Spanish SentiStrength? and propose a simple algorithm for evaluation of sentiment contribution. The work was completed in the framework of the project WAYRA (Telefonica-Peru). The results proved to be promised: opinion analysis of parameterized texts showed the accuracy about 75% with elementary classifier

Keywords: opinion mining, SentiStrength?, Facebook and Twitter

ACM Classification Keywords: I.2.7 Natural Language Processing

Link:

PARAMETERIZATION OF COMMENTS FROM PERUVIAN FACEBOOK AND TWITTER: LEXICAL RESOURCES AND ALGORITHM

Angels Catena, Mikhail Alexandrov, Roque López

http://foibg.com/ibs_isc/ibs-27/ibs-27-p16.pdf