2015 ∙ 11 ∙ 22  /  Work

2000 and Late : Text Mining Twitter

I know. Mining Twitter and making word clouds are hardly thrilling in 2015. Still, being able to gather tweets, run them through text analytics, and show the results with tools that are freely available is really fun.

Beyond having fun, my goal is to make a system for choosegratitude to see how people say thank you, and I am now well on my way.

To begin with, I read a post about how to mine Twitter using tweepy and then ran my tweets through TextBlob. I only wanted to find the top words in the posts, a straightforward task with TextBlob, so getting my data into the right structure was my key challenge. I needed to keep the data in tabular form for the visualization software I use, forcing me to stack the words in a long list, and I needed to keep track of which words belonged to which posts so that I could merge in other metadata at the post level.

Following is a dashboard board showing the top words in the tweets and the country of the person tweeting, where available. You can click on a word or a country to read tweets with that word or from that country.