Sentiment Time Series Analysis on US Economic News

Abstract

In the internet world we will have a lot of information, public opinions, and researchers' comments on the economy of a country, but it will be very difficult to analyze these opinions. Analysis of these opinions is very important to know how the economy of a country does change and to predict the economy of the country. Sentiment analysis does analysis of public opinion in the textual form, and it provides either of positive, neutral or negative sentiments of the textual comment given on the economy. Sometimes, sentiment analysis may not develop a model for better prediction and judgement of public opinion. In this chapter, the authors proposed a method that integrates sentiment analysis with the time series. They proposed a method to create a domain-specific lexicon to calculate the sentiment of the textual opinions over the economic dataset. This chapter implements the sentiment time series model on economic news by using the lexicon-based approach based on built-in lexicons and domain-specific lexicon.