TY - JOUR
AU1 - Rani, Sujata
AU2 - Kumar, Parteek
AB - Sentiment analysis (SA) of natural language text is an important and challenging task for many applications of Natural Language Processing. Till now, researchers have used different types of SA techniques such as lexicon based and machine learning to perform SA for different languages such as English, Chinese. Inspired by the gain in popularity of deep learning models, we conducted experiments using different configuration settings of convolutional neural network (CNN) and performed SA of Hindi movie reviews collected from online newspapers and Web sites. The dataset has been manually annotated by three native speakers of Hindi to prepare it for training of the model. The experiments are conducted using different numbers of convolution layers with varying number and size of filters. The CNN models are trained on 50% of the dataset and tested on remaining 50% of the dataset. For the movie reviews dataset, the results given by our CNN model are compared with traditional ML algorithms and state-of-the-art results. It has been observed that our model is able to achieve better performance than traditional ML approaches and it has achieved an accuracy of 95%.
TI - Deep Learning Based Sentiment Analysis Using Convolution Neural Network
JF - Arabian Journal for Science and Engineering
DO - 10.1007/s13369-018-3500-z
DA - 2018-08-10
UR - https://www.deepdyve.com/lp/springer-journals/deep-learning-based-sentiment-analysis-using-convolution-neural-4Ag8reQPJ8
SP - 3305
EP - 3314
VL - 44
IS - 4
DP - DeepDyve
ER -