Participación de SINAI Word2Vec en TASS 2014
-
1
Universidad de Jaén
info
Publisher: Sociedad Española para el Procesamiento del Lenguaje Natural
Year of publication: 2014
Congress: Sociedad Española para el Procesamiento del Lenguaje Natural. Congreso (30. 2014. Girona)
Type: Conference paper
Abstract
In this paper it is described the participation of the SINAI-word2vec team at the task 1 on polarity classification of the TASS 2014 workshop. Our system uses supervised learning with SVM over the summatory of word vectors in a model generated from the Spanish Wikipedia. Our solution does not follow the vector space model, nor lexical or syntactic analysis is applied. Just considering each word independently represented in a 200-dimensional space of a Word2Vec model, it is possible to capture the semantics of a tweet and report good results in polarity classification.