Abstract ▼ AI-Helper

In the Internet, a sheer number of newspaper articles about economy and finance are generated every day. This paper attempts to predict stock price fluctuation by making use of internet news appearance of companies through a text mining procedure. Therefore, this paper presents a procedure to predic...

In the Internet, a sheer number of newspaper articles about economy and finance are generated every day. This paper attempts to predict stock price fluctuation by making use of internet news appearance of companies through a text mining procedure. Therefore, this paper presents a procedure to predict the future stock price fluctuations where N-gram is used to express words and Word2vec to label words. In contrast with only one time window of news information in other papers, a methodology using word trend data is proposed. SVM (Support Vector Machine) is used to predict the stock price fluctuation. The performance of the proposed procedure is evaluated by using historical news and finance data. The results show that the word trend based methodology gives a improvement while Word2Vec gives approximation to existing method.