Comparative Study of Combination of Convolutional and Recurrent Neural Network for Natural Language Processing

EasyChair Preprint 1451

Abstract

Nowadays social networks are very inclusive and there is a lot of raw information.  Facebook has the most users. Most of the information on Facebook is comments. Because of this reason we choose Facebook for this research. The aim of this research is to comment mining in the Facebook social network. In this research, at the first for removing noise and data cleaning, we apply 10 preprocessing methods on the dataset. Then the data classified by using CNN, LTMS, CNN-LTMS and LTMS-CNN methods. The result showed that the most accuracy belongs to combined LTMS-CNN method and fastest approach is CNN method. We can use these methods for getting useful data on social networks.

Keyphrases: CNN, Comment Mining, LSTM, NLP, text mining

@booklet{EasyChair:1451,
  author    = {Afshin Shirbandi and Babak Moradi},
  title     = {Comparative Study of Combination of Convolutional and Recurrent Neural Network for Natural Language Processing},
  howpublished = {EasyChair Preprint 1451},
  year      = {EasyChair, 2019}}