Information Technology Reference
In-Depth Information
is incrementally increasing day by day, is one of the latest trends in the recent era.
This 140-character-allowing micro-blogging social platform has a wide range of
users varying from people to organizations, such as politicians, celebrities, and com-
panies. According to Kwak et al. [
16
], the number of Twitter users was 40 million
in the world in July 2009, but in August 2014, it is revealed on Twitter's official
web page
1
that Twitter has 271 million monthly active users, although it is a young
company established in 2006. This drastic change in the number of users sheds clear
light on the growth of the company.
The recognition that such a growing company has a vital impact on everyday life
has become an integral element of encouragement for researchers to conduct studies
in regards to the reflections of tweets on the real world to make some predictions.
For example, Asur and Huberman [
3
] were able to forecast box-office revenues for
some movies by using the tweets. Another study showed that Twitter has a vital role
in elections if used effectively. Tumasjan et al. [
31
] discovered that messages in favor
of a candidate party can alter the election result. A similar study by Diakopoulos and
Shamma [
10
] also demonstrated that Twitter is one of the best ways to predict the
election results. In that study, the tweets, which were sent by the users during the
2008 USA presidential debate, were tracked. It was found that the number of negative
tweets posted by the users was less than the number of negative tweets posted when
McCain spoke. Afterward, Obama won the election against McCain. Jansen et al.
[
13
] analyzed more than 150,000 micro-blog posts which contain brand comments,
sentiments, and opinions. They showed that micro-blogging is a kind of electronic
word-of-mouth of customers which are related to brands and products.
As it can be understood from the studies above, Twitter has become one of the
best ways of getting customers' opinions and making predictions about the results
of elections and events. Considering all the predictions made in this way, several
precautions can be taken in case of an unfavorable outcome. For example, a company
can decrease the prices by tracking the sentiment of tweets about a particular product.
A negative trend on twitter can lead to a decrease in prices. In accordance with this
kind of purposes, in this chapter, a web-based system was created in order to extract
information from Twitter by tracking sentiment.
In this chapter, the primary objective is to present a web-based Twitter sentiment
tracking tool. This tool collects the tweets about four brands namely, Facebook,
Twitter, Apple, andMicrosoft, in an hourly basis in 17 Anglophone cities fromwhere
these tweets were sent. The list of the cities used in this analysis can be observed
in Fig.
2.1
. After collecting tweets, the system analyzes sentiments of tweets and
classifies them as positive or negative by using two classifier methods namely Naïve
Bayes and Maximum Entropy. Later on, the system determines the winner brand of
each city according to the percentage of positive tweets by using the information
coming from the users located in selected cities. At the end, the winner brands can
be seen using Google maps. For example, if the winner brand is Microsoft in New
York on a selected day, the system used in this chapter shows the Microsoft logo
Search WWH ::
Custom Search