Evaluation of potential Spanish text markers on social posts asfeatures for polarity classification
Archivos
Fecha
2017
Tipo
artículo original
Autores
Casasola Murillo, Edgar
Leoni de León, Jorge Antonio
Marín Raventós, Gabriela
Título de la revista
ISSN de la revista
Título del volumen
Editor
Resumen
This work describes the identification and evaluation process of potential text markers for sen-timent analysis. Evaluation of the markers and its use as part of the feature extraction processfrom plain text that is needed for sentiment analysis is presented. Evaluation of text markerobtained as a result of systematic analysis from a corpus over a second one allowed us to iden-tify that emphasized positive words are strong indicators for positive text. The second corpusallowed us to evaluate the relation between the polarity of emphasized words and the text theyappear in. Evaluation of the markers for polarity detection task in combination with a polarizeddictionary produced polarity classification average precision of 56% using only three markers.This are promising results compared to the top 69% obtained using more features and specializeddictionaries for the same task
Descripción
Palabras clave
Information retrieval, Natural language processing, Sentiment analysis, Feature vectors, Text classification