Concatenate text embeddings for text classification

Text embedding has gained a lot of interests in text classification area. This paper investigates the popular neural document embedding method Paragraph Vector as a source of evidence in document ranking. We focus on the effects of combining knowledge-based with knowledge-free document embeddings for text classification task. We concatenate these two representations so that the classification can be done more accurately. The results of our experiments show that this approach achieves better performances on a popular dataset.