Google Analytics Code

Thursday, April 15, 2010

SAS Seeks to Improve Data Mining of Social Media - Bits Blog - NYTimes.com

  • There is rich data available in consumers' social media streams but how can it be collected, sorted, and meaning made from it? SAS Social Media Analytics seeks to do just that. The software will automate sentiment analysis for social media data. One tester said the software achieved 92% accuracy with that of human readers. 

    tags: research, SAS, smm, Twitter, sentiment analysis, example

    • No one doubts that social media – all the stuff on Facebook, Twitter and other online forums – provides a rich lode of user sentiment that companies ought to be able to exploit. And not just to sharpen their marketing, but also to improve their products and services – potentially, the ultimate source of customer views and a crowd-sourced suggestion box.

      But the challenge is how to make sense of the ocean of information, find meaningful patterns and use it as guide for action. The digital tools for social media monitoring and analysis, analysts say, are still primitive.

    • SAS Institute, the leader in advanced business intelligence and data analytics software, thinks it can do better. It is introducing a software service on Monday called SAS Social Media Analytics that analysts say seems to represent a step ahead in social media analysis tools.

      The new SAS service is powered by the natural-language processing technology that SAS picked up in 2008, when it acquired Teragram, a Cambridge, Mass., specialist in that niche of artificial intelligence.

    • Until now, they say, software alone has not come near the caliber of “human readers” (a person reading through thousands of tweets related to a given company and rating each as positive or negative).
    • Katie Delahaye Paine, head of a communications measurement and research firm in Berlin, N.H., said the best sentiment analysis software has been about 70 percent accurate, compared with human readers. But when she tested the SAS service it scored 92 percent.
    • In the case of a hotel company, Mr. Chaves said, the technology could be used to find people on Twitter who have the highest percentage of tweets related to the hotel industry or that specific company. Then, a heavy hotel twitterer would be graded by number of followers, a sign of influence. Next, the hotel twitterer would be graded by how often he or she is retweeted, an indication of the engagement of the poster’s online audience.
    • Along with online ads, he noted, working with “these influencers can become an important part of a company’s digital media mix.”

Posted from Diigo. The rest of my favorite links are here.

No comments: