Enhanced Analytics and Sentiment Analysis Arrive in Social Analytics


We are proud to announce an update to the Microsoft Codename “Social Analytics” Lab.  This update builds on the services we announced in October by providing the following new capabilities:

  • Enhanced analytics in our API
  • Sample usage of the enhanced analytics in the Engagement Client – 5 new analytic widgets
  • Improved sentiment analysis for Tweets

1. Enhanced Analytics:

Our existing API now provides richer analytics views of social data.  This makes it easier to find top conversations, contributors, keywords and sources related to any filter with a simple ODATA query and returns conversations related to these calculations to your application.

Here’s a sample* ODATA query returning the top 5 keywords found in a filter for the past 7 days:

https://api.datamarket.azure.com/Vancouver/VancouverBillGates/GetTopKeywords?TimeIntervalSize=’day’&TimeIntervalCount=7&TimeIntervalEnd=&FilterId=4&ResultsLimitTo=5

You can start investigating the details of the API by going to this link.  If you are new to Social Analytics, check out our API overview here.

* Note: You will need to use your existing credentials for the lab to use this sample code. (Look at the API section here for details)

2. New Analytics Widgets in the Engagement Client:

See the new enhanced analytics in use in our sample Silverlight UI – the Engagement Client.   We built these 5 analytic widgets using the new analytic enhancements to our API.   These widgets provide the following analytic views for any filter:

  1. Top Conversations
  2. Top Contributors
  3. Top Keywords
  4. Volume
  5. Sources

Here’s an illustration of the new analytics in the sample Silverlight client:

    3. Improved Sentiment Analysis for Tweets

    In this release, we snuck in a little bonus!  We updated the sentiment engine with minor enhancements to improve the analysis of sentiment in Tweets.  

    That’s what’s new in this release!   In a few short months, we’ve made major progress on this experimental cloud service.   

    Now it’s time for you to try the updated lab and let us know what you think.  We’re already working on the top feature request from our fall release – giving you the ability to define your own topics.  We plan on making that functionality part of a future release.

    For those new to this Lab, please check out our initial blog post for an overview of Microsoft Codename “Social Analytics”.     Here are some additional links from that post for your convenience:

    Comments (5)

    1. rogerj says:

      How about more details for the upgrade to the sentiment engine, as well as a description of the technology/algorithm(s) behind it?

      Cheers,

      –rj

    2. Hello Roger,  welcome back and thanks for your continued interest in our incubation project!   We are using an upgrade to the Microsoft Research sentiment code we have been using in prior releases of our project.   Stay tuned to our blog for our next blog post with more details on this very topic!

    3. Fernanda says:

      Hi! I was wondering what is the timeline for the new release that will include the ability of defining your own topic. Do you have an estimate of when it may become available? thanks much!

    4. Fernanda, we are not able to share the timeline for the ability to define your own topic at this time.   When we are able to share that information, we will!  

    5. Alberto says:

      Text analysis does work only on english?