Detección de polaridad por tópicos para textos cortos en español
Fecha
2019-07-20
Autores
Ramón Hernández, Alejandro Ariel
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
Las opiniones son importantes para la toma de decisiones, de ahí la necesidad de analizar variantes para extraer mayor cantidad de información de las mismas. Entre la información relevante que ofrecen, la polaridad que expresa y los tópicos a los que se refieren constituyen aspectos a trabajar. Generalmente la polaridad se calcula a nivel de documento o de oración, por lo que en ocasiones este análisis no brinda información suficiente para la toma de decisiones. En este trabajo se propone un procedimiento para calcular la polaridad de las opiniones agrupadas por tópicos. Para esto un método de agrupamiento del tipo jerárquico aglomerativo se aplica para obtener grupos por tópicos y la polaridad se calcula con el uso del recurso SpanishSentiWordnet. El procedimiento permite conocer el grado de positividad y negatividad expresado sobre los distintos tópicos tratados en un conjunto de opiniones.
Opinions are important for decision making, hence the need to analyze variants to extract more information from them. Among the relevant information they offer, the polarity that it expresses and the topics to which they refer constitute aspects to be worked on. Polarity is usually calculated at the document or sentence level, so sometimes this analysis does not provide enough information for decision making. This paper proposes a procedure to calculate the polarity of opinions grouped by topics. For this, a clustering method of the agglomerative hierarchical type is applied to obtain groups by topics and the polarity is calculated with the use of the SpanishSentiWordnet resource. The procedure allows knowing the degree of positivity and negativity expressed on the different topics treated in a set of opinions.
Opinions are important for decision making, hence the need to analyze variants to extract more information from them. Among the relevant information they offer, the polarity that it expresses and the topics to which they refer constitute aspects to be worked on. Polarity is usually calculated at the document or sentence level, so sometimes this analysis does not provide enough information for decision making. This paper proposes a procedure to calculate the polarity of opinions grouped by topics. For this, a clustering method of the agglomerative hierarchical type is applied to obtain groups by topics and the polarity is calculated with the use of the SpanishSentiWordnet resource. The procedure allows knowing the degree of positivity and negativity expressed on the different topics treated in a set of opinions.
Descripción
Palabras clave
Detección de Polaridad, Textos Cortos en Español