A Neural Network Model for Syntactic and Semantic Disambiguation

JOSÉP SOPENA , MARTA ALEGRE , JOAN LÓPEZ , AGUSTI LLOBERAS

Resumen


IN THIS ARTICLE, A NEURAL PARSER IS DESCRIBED THAT COMPUTES SENTENCE STRUCTURE AND ACHIEVES COMPOSITIONALITY IN A SIMPLE AND EFFECTIVE WAY. THE MODEL IS COMPOSITIONAL IN THE SENSE THAT IT IS ABLE TO PARSE NEW STRUCTURES -NEVER HAVING BEEN SEEN BEFORE- WHICH ARE RECURSIVE COMBINATIONS OF KNOWN STRUCTURES. THE SYSTEM THAT WE PROPOSE IS MADE UP OF TWO MODULES. THE FIRST MODULE, THE ANNP, IS THE ONE THAT ACTUALLY DOES PARSING. TO TEST THE EFFICIENCY OF THIS MODEL WE RAN EXPERIMENTS USING SMALLER TRAINING SETS AND REAL TEXTS SETS. THE RESULTS WERE VERY PROMISING. THE SECOND MODULE, ANNSP, DOES THE SYNTACTIC AND SEMANTIC DISAMBIGUATION (WSD) FOR REAL WORL TEXTS. WE PRESENT RESULTS COMPARING PERFORMANCES OF THE NEURAL APPROACHES ON SEVERAL WELL STUDIED DISAMBIGUATION TASKS. IN ALL CASES WE SHOW THE NEURAL APPROACH EITHER OUTPERFORMS OTHER METHODS OR PERFORMS COMPARABLY TO THE BEST OF THEM.

Palabras clave


;NATURAL NETWORKS; NATURAL LANGUAGE; PARSING; SYNTATIC DISAMBIGUATION; SEMANTIC DISAMBIGUATION

Texto completo:

pdf


Contacto:
Oscar Zavala