As Twitter allows users to search for tweets with search
terms, it is common for researchers to construct a corpus of
tweets relevant to their own needs. A huge number of tweets
can be collected as there are approximately 5,700 new tweets
posted to the site every second1. Tweets are tagged with
metadata such as the language spoken by the user, the user's
geographical location, and date. This provides opportunities
to target analysis to a specic language, topic or current
event [9]. In our case, we searched for the emoticons in
Table 1 in order to collect tweets with emotional content.