Máquina de aprendizado extremo aplicada à análise de sentimentos

Abstract

The instantaneous increase the amount of information (and opinions) available on the World Wide Web (Internet), leverages the need for the creation and improvement of methods and tools able to exploit the opinational and emotional contents published, diary, by a growing amount of users. Computational Intelligence techniques have been used in wide Data Mining applications. Out of numerous Artificial Intelligence techniques, Artificial Neural Networks and Support Vector Machines (SVMs) have been played the leading roles. However, it is known that both, Neural Networks and SVMs, face some challenging and troubles issues such as: slowness learning speed, not trivial human intervention and/or poor computational scalability. Extreme Learning Machine (ELM), an emergent method to training Neural Networks, arrives to overcome some of this challenges faced by other techniques. ELM works for Single-Hidden Layer Feedforward Networks (SLFNs) and its essence is that the hidden layer of SLFNs doesn’t need to be tuned. When compared with the traditional computational intelligence techniques, ELM provides better generalization performance at a much faster learning speed without requiring human intervention. This study intends to evaluate these and other features of the recent ELM in order to perform a quantitative and qualitative analysis. The intention is to conduct a comparative between the proposed method and the SVM, KNN e MLP methods, to the analysis of sentiment and opinion mining.