چكيده به لاتين
Abstract
Market and economy are among the important factors of the welfare of a community. Because
understanding the power of market helps us in market future prediction, studying the market
and the manner of its movement is considered as a key issue. The ability to predict the market
is equal to the ability to generate wealth by avoiding losses and gaining more profit. This
prediction provides us with the ability to trade and take profit from this system. Several research
studies have been conducted on the prediction of the market conditions by means of machine
learning methods. In the last decade, text processing methods have been widely applicated by
researchers for the purpose of predicting stock prices. However, unfortunately, in the Persian
language, this approach has not been adopted. In this thesis, by proposing an unsupervised
sentiment analysis algorithm for Persian language and utilizing a self-organized fuzzy neural
network, we propose a stock value prediction algorithm. The experiments results, on the one
hand, prove the absolute superiority of the proposed unsupervised sentiment analysis method
to the state-of-the-art of the Persian unsupervised sentiment analysis, and on the other hand,
prove the superiority of the proposed stock value prediction algorithm over the base method
(i.e. self-organized fuzzy neural network without sentiment input).
Keywords: stock price, sentiment analysis, stock prediction, self-organized fuzzy neural
network, social networks
