A Textual Analysis of Financial Articles for Prediction of Polarity

Author Name(s): Nischal S. Puri , Dr. Avinash J. Agarwal, P.S. Prasad
Author Email: nischalspuri@gmail.com

Abstract

Finding Polarity has been used in various domains. In this approach our domain is Stock market which is a promising domain. [11] Using information such as discussion a thread, blogs of expert in their respective fields, technical data, quarterly results of company etc. has been generated such enormous information is available on internet. Generally information is referred as ‘Unstructured Data’. In this paper we discussed how we can use this unstructured data for determining the polarity of text. In Data mining techniques many applications have used to interpret such data. However, Stock data has dependencies among different parameters, and its characteristics change with time and natural scenario. [11] Here we discussed techniques which can be helpful for predicting the prices of stocks. However, there is only outlined structure for the problem in our interested domain. The discussed process will identify words in an article and compares with the dictionary. This dictionary contains words which will be helpful to find polarity. A weight is assigned to each word in order to calculate the polarity. The calculation of the final sentiment score produced by the model. The application of our technique is openly available in the open-source code

Keywords Textual Analysis; Data Mining; Machine Learning; Bags of Words; Stock Market.

Introduction

In [11] discussed a huge amount of data, company annual or periodic meetings, company’s quarterly results of etc has been available on various Internet websites. Therefore, the important problem is to extract knowledge from these information and make use of information. For this, knowledge about techniques involved in data mining is required. In now days, it has been able to study the huge data because of developments in technology. Here, we discuss the Analysis of extraction of text techniques for trading in stocks to find the rise and fall of stocks before the actual movement occur in order book profit either by increase / decrease in the stock price. In General practice, two popular Prediction Methodologies in domain of Stock are as follows [11]. Fundamental Analysts carried out, by this method is more focused with the basics of the company than the stock. The analysts takes their conclusions based on the past working of the company, the earnings prediction etc. [11] .When put on Machine Learning to Data of Stock, we are more concerned in Technical Analysis to understand if our algorithm can precisely learn the primary patterns in the time series of stock. This thought, Learning can also show a major role in assessing and predicting the performance of the stock and other parallel parameters supportive in Analysis of fundamental .Indeed, the most effective automated prediction of stock and recommendation structures use some hybrid analysis model relating both Technical and Fundamental Analysis

 

Conclusion

The model can be made further strong and adapted in future. In time ahead we attempt to shape a model near and try to uncover other procedures to study the polarity of text. Thus we recommended a basic polarity analysis model in R Language. Polarity analysis is the study of the sentiments and feelings (i.e. thoughts, reactions and opinions) which are articulated in the news information /blog columns /twitter messages etc., using processing of natural language tools. [11] The purpose is to mine data from unstructured and semi-structured data unearthed in these tweets/blogs/articles. Due to usage of computational linguistics, artificial intelligence, and computer science enables formulization of NLP. To find the polarity of text various NLP models can be used for more accuracy. In areas like Finance, Politics, Policy making, Business intelligence polarity analysis is a popular topic now a days, and has accepted widely etc

792 total views, no views today

Download PDF File

About the author: admin