Análisis lingüístico de expresiones negativas en tweets en español
- Esther Villar Rodríguez 1
- Ana García Serrano 2
- Marta González Rodríguez 1
-
1
Tecnalia
info
Tecnalia
Derio, España
-
2
Universidad Nacional de Educación a Distancia
info
- Alberto Díaz Esteban (coord.)
- Iñaki Alegria Loinaz (coord.)
- Julio Villena Román (coord.)
Publisher: Sociedad Española para el Procesamiento del Lenguaje Natural
ISBN: 978-84-695-8349-4
Year of publication: 2013
Pages: 4-11
Congress: Sociedad Española para el Procesamiento del Lenguaje Natural. Congreso (29. 2013. Madrid)
Type: Conference paper
Abstract
This paper presents the analysis performed on the collection TASS of tweets in Spanish, in order to improve the automatic analysis of the opinion, on the basis of new lexical resources and processes to identify the scope and focus of the negation. As in this collection, in addition to the basic classes, positive, negative, neutral and none, it is taken into account two new classes very positive and very negative, it is also studied in detail the cases involving modifiers affecting (increasing or investing) the opinion of the tweets. Throughout the article, we present representative examples of each process and conclude with examples of open problem not yet reflected in this work relating to the irony, the need for context and inherent subjectivity of the tweets, which hinder the correct or at least agreed analysis of the opinion.