Because Splunk can index any kind of data, many of our customers have found it useful for indexing and analyzing social media events like Tweets, Facebook posts, and blog posts.
Tweets posted during Hurricane Sandy from the affected regions were indexed and analyzed. They were used to track how many people left the area and when they left relative to the arrival of the storm, people’s sentiment regarding levels of critical supplies, and people’s levels of anxiety and fear.
Using built in Splunk analytics capabilities combined with add-ons like Sentiment Analysis, this site indexes and correlates data from regulations.gov to better understand public sentiment as it relates to specific regulations. The site provides insight on which agencies get the most comments in which months (hint: the IRS gets a lot of attention in February).
GETTING DATA IN
The best option is to use a social media data aggregator like Datasift. Datasift does the heavy lifting of collecting and aggregating social media events from a wide variety of sources and filtering those events based on your criteria. Splunk has a partnership with Datasift and worked hand-in-hand with them to create a plug-n-play modular input to index those social media events in real time.
Having Datasift do the aggregating and filtering leaves you tons more time to use Splunk’s powerful search and analytics capabilities to get at the answers you need from your data.
If you’re more of a DIYer, you can use Splunk’s REST API Modular Input to access any data source with an API. Here are detailed instructions on using the modular input for Twitter (http://discoveredintelligence.ca/stream-twitter-splunk-10-simple-steps/). But, of course, you can use the Modular Input for any data source.
GETTING ANALYTICS OUT
So, getting data in is all well and good — now how do we get the analytics out?
Of course, Splunk’s core search and analytics capabilities are more than sufficient to filter, correlate, and visualize social media data to identify critical patterns.
There are also a number of apps that provide pre-configured dashboards and other types of analytics.
One of Splunk’s own created a sentiment analysis app that enables you to “train” Splunk to categorize tweets (or any social media event, for that matter). For example, if you train Splunk to categorize social media events by “positive”, “neutral”, and “negative”, you can track sentiment of your brand over time.
Another great app, and a really great place to start with all this, is the App for Twitter Data. This app comes with its own modular input and a collection of dashboards — all you need to do is provide your Twitter credentials.
Hope this is useful! Please write with questions.