Stock Price Trend Prediction and Recommendation using Cognitive Process

Stock Price Trend Prediction and Recommendation using Cognitive Process

Vipul Bag (SGGSEIT, Nanded, India) and U. V. Kulkarni (Computer. Science & Engineering Department, SGGSIET, Nanded, India)
Copyright: © 2017 |Pages: 13
DOI: 10.4018/IJRSDA.2017040103
OnDemand PDF Download:
No Current Special Offers


The paper emphasizes on stock price trend prediction based on the online textual news. Cognitive process uses existing knowledge and generates new knowledge. Contextual features (CF) from news sites are extracted & recommendations based on the interpretations are generated. A Naïve bays classification algorithm is used to classify the news sentiments. A News Sentiment Index (NSI) is calculated and effect of the news on particular stock is calculated to predict the trend. Along with news sentiment index, technical quality of the same stock is calculated by various statistical technical indicators which are called as Stock Technical Index (STI). The weighted index of NSI and STI is used to predict the trend of stock price. In the previous recommendation systems, the context of the recommendation is not considered. However, it is shown in this research that if the authors consider the news context while recommendation, the performance of the recommendation system will drastically improve. The results are compared with traditional systems and it shows significant improvement.
Article Preview

Prediction in stock market is always a hot topic for researchers in data mining. Many researchers have worked on predicting the closest, best and strong rule. Neural Network, Genetic Algorithm, Association, Decision Tree and Fuzzy systems are widely used to predict stock prices. Among all the above techniques text mining is the straightforward technique.

A decision support system is developed based on the recommendations given on different sites and analyzing them in the work done by the authors. (Gottschlich & Hinz, 2014). The work reported in (Tang, Liang, Hu, 2008; Feng, Park, Kim et al., 2009) and others was on market basket analysis to generate association rules to develop decision making system. The work done in the research (Hagenau, Liebmann, Neumann, 2013; Das & Chen, 2007) is on sentiment analysis using text data. However, in the work the technical analysis is not carried out.

Another related work done by (Feng, Yu, Lu et al., 2008], the aim was devising an efficient and flexible approach that recommends appropriate investment types to stock investors by discovering useful rules from past changing patterns of stock prices stored in a database.

In the work carried out by (Paranjape & Deshpande, 2013), have shown that the association rule mining with a support confidence framework can be used to build a stock market portfolio recommender system and their approach demonstrates the application of soft computing techniques like ARM and fuzzy classification in the design of an efficient recommender system.

The work reported by (Tung, Lu, Han et al., 2003) have worked on apriori algorithm and modified it so as to suite the uncertain dataset and named it U-Apriori algorithm. The computational problem of U-Apriori is identified and proposed a data mining framework to address this issue. They proposed the LGS-Trimming technique under the framework and verified, by extensive experiments, that it achieves very high performance gain in terms of both computational cost and I/O cost. The work done in (Goumatianos, Christou & Lindgren, 2013) is limited only for generating intraday patterns. In research papers (Lin, Yang, Song, 2011; Lin, Yang, Song, 2009) echo state network along with technical analysis is used to develop expert system.

Complete Article List

Search this Journal:
Volume 8: 4 Issues (2021): Forthcoming, Available for Pre-Order
Volume 7: 4 Issues (2020): Forthcoming, Available for Pre-Order
Volume 6: 4 Issues (2019): 3 Released, 1 Forthcoming
Volume 5: 4 Issues (2018)
Volume 4: 4 Issues (2017)
Volume 3: 4 Issues (2016)
Volume 2: 2 Issues (2015)
Volume 1: 2 Issues (2014)
View Complete Journal Contents Listing