Chronological Word Frequency Analysis of the 2017 Iran-Iraq Earthquake

摘要

The aim of this paper is to understand the chronological communication behavior of Twitter users immediately after the 2017 Iran-Iraq earthquake by identifying word frequencies and an overall sentiment analysis of their tweets. A total of 20000 tweets were collected over three time periods. In the analysis of the tweets, all retweets were removed. In text mining, a common task is the inspection of word frequencies. The higher the frequency of a word in a time-period the more important that word is in understanding the tweets in that time-period. To quantify the similarity of these sets of frequencies, the Person's product-moment correlation was calculated. The distribution of the word frequencies was plotted for each time-period. To measure how important a word is in a tweet in a particular time-period the term frequency inverse document frequency, tf-idf, was calculated. The findings show that the plotted distributions are typical of a corpus following Zipf's law.