Once more, with feeling, Pats Nation!07 Feb 2017 Tutorial
Rosette API with RapidMiner investigates Twitterverse sentiment surrounding Super Bowl LI
As a Boston area company, you can imagine how proud and excited our staff was at the end of Super Bowl LI on Sunday night! Despite tough competition, the New England Patriots did their job, so today we do ours: helping you extract meaningful information and analysis from your text and social media data.
Sentiment analysis, or opinion mining, is a crucial tool for social media monitoring, brand analysis, trend analysis, and public opinion measurement. In honor of the New England Patriots’ record shattering win on Sunday, we decided to take a look at tweet sentiment using Rosette with Rapidminer Studio, the number one predictive analytics platform on the market. The Rosette text analysis toolkit, including document and entity level sentiment analysis, is available on the Rapidminer marketplace and free to download.
Twitter has some strong feelings about this…
Major sporting events like the Super Bowl or the World Cup regularly generate enormous volumes of social media data, tens of millions of posts on any given platform. For the sake of demonstration, we decided to focus only on Twitter, analyzing the sentiment of a sampling of tweets containing relevant hashtags.
One iconic part of every Super Bowl is the halftime show. We wanted to see if the Twitter-sphere was as excited as we were about Lady Gaga’s performance, and it seems like they were! Based on our sample of 5000 tweets that mentioned #pepsihalftime, the majority of Tweet-ers enjoyed Gaga’s performance!
We did the same with tweets that metioning #patriots and #NEpatriots. These hashtags were associated with slightly more positive than negative sentiment, but not by much!
Although we’re admittedly biased, we tried to be fair by also analyzing tweets mentioning #falcons or #falcon. We suspect the shocking comeback by the Patriots in the 4th quarter led to the high volume of negative tweets directed at Atlanta.
Try it yourself!
It’s easy to build your own Twitter sentiment analysis model with Rosette for RapidMiner. Download the Rosette Text toolkit on the RapidMiner marketplace and make sure you have an active Rosette API key.
If you don’t have a key yet, you can sign up for free on our developer site, no credit card required. In our process we used three operators: “Search Tweets” to collect data, “Remove Duplicates” to clean it, and Rosette’s “Analyze Sentiment” to evaluate the sentiment itself.
To get a broader picture of public opinion surrounding the Super Bowl, you could try analyzing tweets that mention other hashtags, like #SuperBowlLI, #LadyGaga, or #RogerGoodell (though we suspect we can predict how people feel about that last one…). If you come up with anything interesting, let us know! We love to feature user stories on our blog!
If you’re new to RapidMiner and need some help getting started, check out our Quick Start Guide to Extracting Entities in RapidMiner Studio with Rosette!