Stock prediction method and apparatus by ananyzing news article by artificial neural network model

摘要

The present invention relates to a method and apparatus for predicting a stock index by analyzing a news article using an artificial neural network model, and more specifically, inputting a number of news articles into a machine-learned artificial neural network model, and daily from this model. It relates to a method for predicting the stock price index and the apparatus for performing the method. According to the present invention, a relation unit (O1; P; O2) having a structure similar to the subject (S), the verb (V), and the object (O) is not extracted from the news article, but is lower than this. By extracting word-level information, it provides an artificial neural network with input that reduces information loss. Since the news article title is relatively short and the variation in the number of words used is small, the loss of information is much reduced and useful results are obtained by using word-level information as input without applying the form of relation tuple. In addition, by using a recurrent neural network (RNN) or a long short-term memory (LSTM) model, which is a type of RNN, as an artificial neural network, processing of data appearing sequentially, such as words in news articles, is more accurate, so that the stock index is more reliable. So that you can get a good result.