Complaint Classification using Word2Vec Model

Attempt has been made to develop a versatile, universal complaint grievance segregator by classifying orally acknowledged grievancesinto one of the predefined categories. The oral complaints are first converted to text and then each word is represented by a vector usingword2vec. Each grievance is represented by a single vector using Gated Recurrent Unit (GRU) that implements the hidden state of Recurrent Neural Network (RNN) model. The popular Multi-Layer Perceptron (MLP) has been used as the classifier to identify the categories.