Reconhecimento de expressões de emoções na interação mediada por computador

AUTOR(ES)
DATA DE PUBLICAÇÃO

2008

RESUMO

In the area of Natural Language Processing (NLP) the recent advances in text automatic analysis have led to the emergence of a new area which is the management and recognition subjective aspects in texts, considering factors such as opinion, sentiments, emotions and affection. Among the applications of this area we enumerate: to determine the client satisfaction level, to recognize the difference between opinion emission and description of facts, to recognize entrepreneur profile and interpersonal posture, to identify the possibility of depressive states in people, among other possibilities related to subjectivity. The research in this area requires the development of methods, resources and techniques, that, integrated, can make computational systems able to manage the affective meaning in discourse. However, since these studies are recent, the resources are scarce for the portuguese language. In this way, this dissertation proposes the development of an affective lexicon for the portuguese language, named WordNet Affect BR. This database was developed as a resource for applications that demand a vocabulary of emotion, as a initial step to the study of the recognition of subjectivity in portuguese language discourse. Besides the development of the database, this work presented a chat tool, named Emoticon which was developed in order to validate the content of the WordNet Affect BR database as a linguistic resource proper for the recognition of words related to emotions

ASSUNTO(S)

textos sistemas computacionais léxico afetivo análise automática linguagem natural língua portuguesa ciencia da computacao

ACESSO AO ARTIGO

Documentos Relacionados