Real-time. Social Sentiment Analysis. Twitter. Cloud. Insights. We have your Big Data buzzwords here!
Everyone seems to want to incorporate social sentiment into their business analysis. Well we have the demo for you! Use it for a quick demonstration of what can be done and when the excitement goes through the roof, use it to inspire your own design!
Real-Time Processing – Instant Insights!
First, use Event Driven Processing (EDP) to show the data on a dashboard. In the demo you’ll use StreamInsight, also referred to as Complex Event Processing (CEP), though you may want to use Microsoft Open Technologies’s RX / Reactive Extensions in your own project. Use the dashboard to make real-time decisions and take immediate action. For example, configure your EDP to “pop” only on terms related to your company and your marketing analyst can watch how the volume of tweets and the sentiment (measured positive/neutral/negative in this example) change in response to your Super Bowl ad or a mention of your company on a the news show. She can respond instantly with changes to your website, your own tweets, sales/promotions, or whatever is appropriate to your business.
Data Storage – Enable Insights!
EDP reads the data as it is pushed through a query, there is no inherent storage involved. You could just discard the data and never store it. In this example we chose to store the data for later trending and historical analysis. We take the tweet id, the date/time the tweet was captured, the sentiment score calculated during the real-time processing, and the keyword that caught our attention and store it in SQL Azure. This data is available to other applications that need to join it with existing data and requires fast responses to individual queries. The remaining data including the raw tweet, any geographic data the tweeter chose to share, and other data is dropped in an Azure Blob Store.
Trends, Patterns, and Historical Insights!
Now point HDInsight (Hadoop) to the Azure Blob Storage using HDInsight’s ASV extension to HDFS. You can spin up a Hadoop cluster in Azure, pay for as many nodes as you need for as long as you need them then spin them down to save money. The data remains in the blob store – available for future Hadoop clusters, other applications, archival, or whatever you need it for. Add structure to the JSON data with Hive and now you have rows and columns that can be accessed by BI tools!
Visualization for Powerful Insights!
Now create a PowerPivot tabular model (self-service BI / client side) or an Analysis Service tabular model (corporate BI) to store the relevant data in a highly compressed, in-memory format for fast access. Add in a few Power View visualizations mashing up data from multiple structured and unstructured sources and you can show your business decision makers easily digestible and understandable data in a format they just get and love! Make some decisions, take some actions, and you’ve just shown how to turn free Twitter data into a valuable resource that can have a direct impact on your company!
How to Get These Insights
Follow the instructions on the CodePlex site for the Big Data Twitter Demo project. Set it up, run through the demo, get excited, and go improve your business!
Demo Created By:
Aviad Ezra @aviade_pro
Brad Sarsfield @Bradoop
Later Additions By:
Lara Rubbelke @SQLGal | http://sqlblog.com/blogs/lara_rubbelke/
Robert Bruckner http://blogs.msdn.com/b/robertbruckner/
Cindy Gross @SQLCindy | http://blogs.msdn.com/cindygross