Convolutional Neural Networks and Long Short-Term Memory Networks for Textual Classification of Information Access Requests
Keywords:Convolution Neural Network, Long Short-Term Memory, Government Data, Text Classification
The e-sic system aims to centralize requests for access to information addressed to the Brazilian Federal Executive. However, the volume of received requests may be an impediment to opportunely responses to such requests. The purpose of this paper is to create an automatic classifier of these requests. For this, Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) architectures were analyzed, as well as a combination of these two architectures. The study concluded that the most appropriate network was CNN, because, despite the models tested presented very similar accuracy values, the CNN network took less time to training.