Character-level-Convolutional-Networks-for-Text-Classification

Text classification has become a very important part of Natural Language Processing. It is often used for Sentiment Analysis and Identification of harmful messages on social media networks such as Twitter and Facebook. Achieving this aim is quite difficult hence the numerous researches on going in this subject area. This project is a final semester project for SDS 203 Deep Learning at IMT - Atlantique, based on an article by Zhang et al on Character-level Convolutional Networks for Text Classification , where convolutional networks were used to extract information from text data. Application of CNN to text is often done at the word level where words are vectorised in order for them to be fed into a neural network for training. In this article, Zhang et al propose vectorisation of text at the character level instead of the word level. This allows the CNN to gain more insight about the data. Also,this CNNN does not require any prior knowledge of the words used to train the networks.
The implementation of this project follows a CRISP-DM Methodology for data mining.