Sentiment data from Twitter and Google Trends

Great day of class, last Wednesday. It deserves a short summary, which I am publishing here. The basic idea for the class was the exploration of sentiments around a place, using Twitter.

First I showed my students that Twitter admits a pretty advanced set of search options. Among the most useful is the Geocode options, which gives the coordinates for a certain geographic point. Then Twitter will search for all Tweets that were issued from that geolocation. If you need to know your geolocation, just ask https://whereamirightnow.com. If you want to know the coordinates of any place on Earth, just ask Google Maps. At this moment I am at a computing lab, here:

Latitude:18.466334°
Longitude:-66.105722°

Then, to do the search it’s enough to go to search.twitter.com and enter the following:

geocode:18.466334,-66.105722,0.5km

The third parameter represents the radius of the search circle around the indicated location, in this case 0.5 km.

This way, one can search whatever idea on Twitter around a given place. For instance, we could query all tweets containing the words “Sagrado” or “Universidad” or “estudiar” around those coordinates; and we can even exclude retweets. Like this:

-RT Sagrado OR Universidad
OR estudiar geocode:18.466334,-66.105722,0.5km

Try it here.

Twitter has got also a powerful API that can be used programmatically, so one can design a pretty great program that do sentiment analysis in Python and see what happens!

Of course, I showed students where all this savvy comes from! Here: http://thoughtfaucet.com/search-twitter-by-location/examples/

The Twitter search brought us later to Twistori, the wonderful tool that shows all Tweets in the world displaying a certain word in their text: love, hate, wish, think, etc. Like in the following video:

I love Twistori, and it also comes with a Mac screensaver!

 

After Twistori, a stop at Google Trends was de rigueur. We discussed how to get data on epidemics, or brands, or political issues. G-Trends display great graphs of the data, too.

<<Examples>>

This is the Google Trends search for zika, chikungunya and dengue viruses. We discussed how, exactly as with the Twitter data, one can query and analyze data that may shed light on possible epidemics. And, if we project the term “epidemic” to something that is socially spreading in the same fashion as a biological infection, we can have interesting results by geo-limiting the search to a specific region, all of which is easily done with Google Trends.

And this is the world map corresponding to the same data.

Note that oddity that Serbia, of all European nations, is seemingly very much interested in zika.

Last thing I did was to  inform student we have an interview set with the great Dolors Reig for next Monday, May 2nd at 9:40am, live from Barcelona! Dolors is the champion of the famous blogsite El Caparazón and she tweets as @dreig.


Also published on Medium.

About Antonio Vantaggiato

Professor, web2.0 enthusiast, and didactic chef.
This entry was posted in abierto, information, twitter and tagged , , , , , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *