Text Classification Using Word Embeddings

Growth in the internet made immense impact on the data generation. Most of the data in the world is in textual format. There is need to access and use this data efficiently and easily hence, text classification is widely studied problem in research community. The applications of text classification are also in diverse domains such as news filtering, opinion mining, information retrieval and so on. At core of these applications is machine learning algorithms which need input of fixed length as a vector. Bag-of-Words is popular model used to represent text in numeric vector format. But it has several disadvantages such as word order is ignored, high dimensionality and sparse representation if vocabulary is large. Word embeddings are distributed vector representation of words. These representation inherit semantic notion of similarity also they have shown state-of-art results in many core natural language processing tasks. In presented work, we present text classification using these word embeddings and measure the performance.