Impact of convolutional neural network and FastText embedding on text classification

Abstract Efficient word representation techniques (word embeddings) with modern machine learning models have shown reasonable improvement on automatic text classification tasks. However, the effectiveness of such has not been evaluated yet in terms insufficient vector for training. Convolutional Neural Network achieved significant results pattern recognition, image analysis, and classification. This study investigates application CNN model problems by experimentation analysis. We trained our a prominent embedding generation model, Fast Text publically available datasets, six benchmark datasets including Ag News, Amazon Full Polarity, Yahoo Question Answer, Yelp Full, Polarity. Furthermore, proposed tested Twitter US airlines non-benchmark dataset as well. The analysis indicates that using is very promising approach.