Predicting Stock Price Movement based on Communication Network and Sentiment Analysis

Abstract

Stock market is responsible for trading the shares of public listed companies. Stock exchange facilitates stockbrokers to trade company stocks and other securities. Stock price prediction is one of the most challenging issue which is attracting researchers from many fields including economics, history, finance, and mathematics and computer science. It is difficult to apply simple time-series or regression techniques on stock market because of the volatile nature. Proposed framework attempts to predict whether a stock price sometimes in the future will be higher or lower than it is on a given day. We find a little predictive ability in the short run but definite predictive ability in the long run. Using the social communication network within company among employees, the proposed algorithm can analyze the relationship between communication context and the movements (high and low) of stock price. We have also extended the system by using sentiment analysis for email content which determines whether email context is negative or positive. System gives aggregated result from number of mail exchanged and sentiment of message body.

Microblog Emotion Analysis Using Improved DBN Under Spark Platform

“…The training samples are read from HDFS, and the filter() operator is combined with the stop word list to remove the stop words. The word frequency is recorded using the reduceByKey() operator, while the cache() operator is used to cache the wordRDD and realize parallel pre-processing (Shiral & Torvi, 2020).…”