Twitter Sentiment Tracking for Predicting Marketing Trends - Smart Information Systems: Computational Intelligence for Real-Life Applications - page 59

Fig. 2.3 Flowchart of tweet collecting steps
Relevant tweets that match a query are returned in the JavaScript Object Nota-
tion (JSON) format by the Twitter API. As described on JSON web page, 7 JSON's
properties, which make JSON an ideal data-interchange language, are listed as the
followings: a very lightweight data-interchange format, easy for humans to read and
write as well as easy for machines to parse and generate based on a subset of the
JavaScript Programming Language, JSON is a text format that is not dependent upon
language at all, but it uses conventions that are well-known to programmers of the
C-family of languages.
2.3.3.2 Sentiment Analysis
In this chapter, the Naïve Bayes classification method and Maximum Entropy clas-
sification methods are used to make the sentiment analysis. A Python script was
written in order to classify the tweets collected. Basically, this script first trains the
Maximum Entropy classifier and the Naïve Bayes classifier with training-modified
7 http://www.json.org/ .