Reconhecimento de padrões de orações passivas por meio de Mapas Auto-organizáveis

Abstract

The aim of the current article is to lay the foundations for the development of a descriptive model capable of differentiating several configurations embodied by passive clauses formed with auxiliary verb ser (to be) + past participle of the main verb in specific interaction contexts, based on using the Neural Network of Self-Organizing Maps (SOM Network). All 220 constitutive clauses forming the herein analyzed corpus were allocated by the SOM Network in 46 neurons, which represent different samplesof passive clauses, by concomitantly taking into consideration the classes of all eight parameters selected in the current research. These 46 samples reflect variations in the grammatical and pragmatic-discursive configurations of passive clauses, in a quite thorough manner, by taking into consideration the theoretical apparatus of Systemic-Functional Grammar. The definition of the optimal number of groups allowed us to deduce data from three important groupings that have portrayed the major meanings elicited by the analyzed passive clauses within the scope of all 102 analyzed journalistic news.

Keywords

• passive clauses
• systemic-functional grammar
• neural network of self-organizing maps