Text Classification using Neural Networks

Author: Elene Gabelaia
Keywords: Classification, Deep Learning, Convolutional, Neural Networks, Support Vector Machines, Naive Bayes
Annotation:

It is estimated that about 80% of all information is unstructured, and text is one of the most common types of unstructured data. Because of the messy nature of text, analyzing, understanding, organizing, and sorting textual data is difficult and time-consuming, so we often fail to use it to its full potential. This is where text classification with machine learning comes in. Using text classifiers, we can automatically structure all kinds of text, including wine reviews. This master's thesis represents the classification of wine reviews according to their varieties. For text classification, we used two quite popular algorithms: Naive Bayes, Support Vector Machines (SVM), and the results obtained by these classifiers were compared with the results given by an alternative deep learning model. The approximate accuracy of Naive Bayes and SVM was 63.19% and 80.27% respectively, while the approximate accuracy of the deep learning model we built was: 76.46%. This means that our neural network competes quite well with some of the more common text classification methods.