Abstract:
Due to explosive evolution and popularity of electronic media, Online shopping and
Social media sites, vast amount of user review and experience available in the form of
raw data. It can be used for opining mining or sentiment mining and other pattern
identi cation tasks. Opining mining or sentiment mining and summerization of
review regarding any particular topic used to provide insights and can be used as
feedback to improve or address concerns regarding that topic and helpful in future
planning. Most of the work done so far in this eld foced on run of the mill, well
de ned techniques like K-NN, SVM and others machnine learning algorithms to
classify the text into two or more classes. However, traditional techniques peak out,
in term of accuracy in certain limit. Additional improvement in term of accuracy
reported using deep learning model LSTM-RNN with pre-trained word embedding.
The aim of the present work is to improve existing techniques for opinion mining or
sentiment analysis.
