Contribute Media
A thank you to everyone who has made this possible: Read More

Increasing the trading prediction by mining aggregated human texting messages


A novel approach is proposed to predict intraday directional-movements of the stock in the trading market based on the text of breaking financial news or social media by event data. This work is an effort to put more emphasis on the text-mining methods and tackle some specific aspects thereof that are weak in previous works. The research is a specific market, namely, the trading and stock market,


Sentiment analysis This step is observed the time series of the company‘s event based on the financial news or the social media, which analyses problem of coreference in text mining that is sentiment classification. Fasttext is inspired method that focuses on the meaning of words, which solves to obscure by ambiguity and play on words. This way can classify polarity, subjectivity and intensity in the corpus. According to the results of the classification, then it changes to the word representation. Thus, prediction accuracy increases significantly at this layer that is attributed to appropriate noise-reduction from the feature-space. Model creation and prediction The first layer is termed Sentiment Integration Layer by the deep neural network, which integrates sentiment analysis capability into the algorithm by proposing a sentiment weight that reflects their sentiment. Additionally, this layer reduces the dimensions by eliminating value in terms of sentiment and thereby improves prediction accuracy. The second layer encompasses a deep neural network model creation algorithm. It updates the models with the most recent information available, the deep learning algorithm is extensively evaluated using real market data and news content across multiple years, which is a good fitting capability to evaluate and select the financial news. The third layer is that it uses the recursive neural network explore patterns and of financial news through the recursive and pooling operation.


Improve this page