Text Mining For Stock Movement Predictions: A Malaysian Perspective
Free (open access)
This paper describes a proposed system that extracts keyphrases from online news articles for stock movement predictions. The proposed system is implemented and tested on selected active sectors from Bursa Malaysia, a Malaysian Stock Exchange. The implementation is based on two crucial factors: firstly, stock related news articles are influential in the sense that they can influence a buyer’s attitude and subsequently can cause the stocks to move; and secondly, textual data are considered more superior and contain more information than numeric data (Cho et al. ) because the former not only allows us to predict future stock prices but at the same time provides us with reasons as to why it is so. This paper reviewed investigations on how the online Malaysian news articles are mined to extract appropriate keyphrases; which are then subsequently used to predict the movements of stock prices in Bursa Malaysia. The main focus of this paper is to implement appropriate text mining techniques to extract the said keyphrases from online news sources. The proposed system is then trained and tested based on a collection of Malaysian news articles from The Star Online: Malaysia Business News . The findings indicate that the implementation is capable of providing valuable keyphrases whether or not original keyphrases are provided. Keywords: text mining, data mining, stock movement predictions, Malaysia business news, Bursa Malaysia, Malaysian Stock Exchange, keyphrase extraction, KEA. 1 Introduction In recent years, news articles are widely published on the web. Of all these news articles, a large number of them are related to economic and financial concerns
text mining, data mining, stock movement predictions, Malaysia business news, Bursa Malaysia, Malaysian Stock Exchange, keyphrase extraction, KEA.