Numbers aren’t all that analytical competitors are crunching these days. We all know that there is great value in the comments on blogs, Twitter, Facebook, Yelp, and other social sites but mining “sentiment data” isn’t easy.
“Translating the slippery stuff of human language into binary values will always be an imperfect science,” according to the New York Times. “’Sentiments are very different from conventional facts,” said Seth Grimes, the founder of the suburban Maryland consulting firm Alta Plana, who points to the many cultural factors and linguistic nuances that make it difficult to turn a string of written text into a simple pro or con sentiment. “ ‘Sinful’ is a good thing when applied to chocolate cake,’ he said.”
Yet, despite the difficulty, the challenge must be met. Companies like Attensity, Clarabridge, Jodange and Scout Labs as well as analytics stalwarts like SAS are working hard to provide the computing savvy and firepower to turn tweets and posts into analyzable data.
“Sentiment analysis is not simply the problem of determining whether a document, a paragraph or even a sentence expresses a positive or negative sentiment or opinion. It is also about entities. Without such information, any sentiment is of little practical use. So one should not only talk about sentiment analysis of documents, paragraphs or sentences, but also about the entities that sentiments have been expressed upon. Here an entity can be a product, service, person, organisation, event or topic,” said Bing Liu, a professor at Illinois University, in Textanalyticsnews.com.
Of course the additional challenge is marrying unstructured with structured data for a full picture of attitudes, sentiments, desires, and intentions.
Customer activity isn’t the only use of text analytics. As reported on the SAS blog, Dr. Colleen McCue has examined how handwritten police notes and data taken from phone calls can be analyzed to predict future locations and potential criminal events. You can learn more from this archived Webcast.
What is your experience with analyzing unstructured or sentiment data — the good, the bad, and the ugly? If there is one improvement you’d make in the tools to gather and analyze it, what would it be?
