Stock Market Data with Deep Neural Networks using Twitter Analysis and Rumour Identification

Abstract

In this current era, the stock market is one of the major businesses which is available online. So, for this social media plays a significant role. Increasing the percentage of people using social media for all types of information will definitely lead to the spread of rumours. In this paper, Twitter is one of the most important keywords and can be described as undoubtedly it was the most popular biggest platform in social media. Here the paper focuses on stock market data, particularly which was posted on twitter as a tweet. Here we have to identify whether that tweet can be either rumour or non-rumour. Experimenting with various machine learning and deep learning techniques BILSTM is the best method, which is the part of neural networks. By this method aggressively we can detect rumours. In this deep neural networks proposed system effectively we are using Bidirectional Long short term memory layer. In this, we are significantly discussing with four types of layers. In this BILSTM the contextual information is considered and analysis is done in two directions termed as forward and backward directions. To associate the problem of rumour identification BILSTM is the correct approach. BiLSTM already combined with LiSTM which comes under reoccurrence neural networks that work only in a unidirectional way. In this BiLSTM layer, it effectively addresses the sequences that are hidden in both forward including backwards for obtaining input.