A recurrent neural network is a neural model in which a bidirectional flow of information is present. In other words, while the propagation of signals in feedforward networks takes place only in a continuous manner in one direction, from inputs to outputs, recurrent networks are different. In recurrent networks, this propagation can also derive from a neural layer following the current one, between neurons belonging to the same layer, or even between a neuron and itself. The set of natural language processing techniques, text analysis, and computational linguistics that are used to identify and extract subjective information in written or spoken text sources is called sentiment analysis. In this chapter, a recurrent neural network is used to classify sentiment in movie reviews.
The following topics are covered in...
