®®®® SIIA Público

Título del libro: Optimal Energy Growth In Variable-Density Mixing Layers At High Atwood Number
Título del capítulo: Recherche et utilisation d'entités nommées conceptuelles dans une tâche de catégorisation

Autores UNAM:
JUAN MANUEL TORRES MORENO;
Autores externos:

Idioma:
Inglés
Año de publicación:
2013
Palabras clave:

Automatic categorization; Automatic-learning; Named entity recognition; Opinion mining; Performance; Sentiment analysis


Resumen:

The researchs presented are part of a text automatic categorization task. Words bearing opinions play an important role in determining the overall direction of the message. But it is essential to identify the elements (targets) which they are intended to relativize the scope. The analysis can also be conducted in the reverse direction. When a target is detected we need to search polarized terms in the context. A first step in an automatic learning from data will allow us to obtain the most important polarity markers. From this basis, we look for targets that appear most frequently in the vicinity of these opinions markers. Then, we construct a set of pairs (polarity marker, target) to show that relying on these couples we can maintain (or improve) the performance of the classifier. © 2013 Proceedings of TALN. All rights reserved.


Entidades citadas de la UNAM: