®®®® SIIA Público

Título del libro: 2016 Fifteenth Mexican International Conference On Artificial Intelligence (micai): Advances In Artificial Intelligence
Título del capítulo: Non-human Subject as discriminating feature in Opinion Mining and Subjectivity Analysis

Autores UNAM:
OCTAVIO AUGUSTO SANCHEZ VELAZQUEZ;
Autores externos:

Idioma:

Año de publicación:
2016
Palabras clave:

Subjectivity Classification; Syntactic Analysis; Opinion Mining


Resumen:

Opinion Mining and Sentiment Analysis rely on the Subjectivity Analysis. Opinions and sentiments belong to a category known as private states: things that can't be observed by another person different than the one who is experiencing them. In this article, we will show a syntactic way to detect and analyse subjective texts, specifically, opinions. Syntactic subjects were closely studied inside 299 newspaper articles. The subject of 202 subjective statements was analysed with the next characteristics: level of abstraction, volition, humaneness, and the relationship established with a verb inside the statement. The previous process was made in order to find patterns in the constitution of the subject in subjective statements. As a result of the analysis, we could notice that non-human subjects increased the probability of the statement to be a subjective one. This information can be used to improve the detection of subjectivity in a text


Entidades citadas de la UNAM: