Author: Linda Rosencrance
Spotfire Blogging Team
Author: Linda Rosencrance
Spotfire Blogging Team
Typically, industry analysts lump sentiment analysis and text analytics together, particularly when they talk about how to find value from social media conversations.
But is that the right way to view these two technologies? Is sentiment analysis a component of text analytics or is it an application on its own? And either way, what about the human element?
Sentiment analysis involves measuring the attitude of a consumer toward a brand by analyzing comments and suggestions left on social media sites such as blogs and social networks. But rather than just analyzing words, sentiment analysis identifies a customer’s attitude about a brand by analyzing things like context, tone, and emotion.
Recently, analytics visionary Seth Grimes (@sethgrimes) indicated that sentiment analysis draws on, but isn’t a subset of, text analytics. “Strong sentiment analysis relies on semantic analysis — on application of natural-language processing (NLP) techniques to identify sentiment objects (entities, topics, and concepts), opinion holders, and the sentiment, attitudes, and emotions that the opinion holders attach to the sentiment objects. But sentiment can also be inferred without any semantic analysis. For instance, the Consumer Confidence Index (CCI) and the Michigan Consumer Sentiment Index.”
In a post earlier this week, Grimes continued to break down why sentiment analysis doesn’t depend on text analytics.
Grimes says sentiment analysis draws on, but isn’t a subset of, text analytics. Although text analytics can help companies determine how consumers feel by analyzing their words, Grimes says there are many more sentiment sources out there than just text. Not only that, but Grimes says you don’t really even need text analytics to get at sentiment in text.
In Grime’s view, if sentiment analysis is a text analytics subset, then a smile, yelling, an angry gesture, and dwell-time on a Web page wouldn’t mean anything.
“They express mood, attitude, and emotion that are conveyed visually, audibly, and via movement, but they’re non-textual and thus can’t be parsed directly via text analytics,” Grimes says.
When it comes to social media, companies can use customers’ unstructured feedback to analyze their sentiments and feelings because people don’t control their emotions as much when they interact in social environments. That way businesses can try to understand consumers’ real feelings behind the feedback they leave and use it to achieve their goals.
“Text analytics is great stuff, but it’s not the be-all and end-all of sentiment analysis,” Grimes says.
One of the many comments responding to Grimes’ post was from Cordell Wise (@cordellwise). He says the problem occurs when companies try to capture all these forms of communication as data.
“I agree there will be inevitable loss of fidelity (meaning) but data is rarely perfect,” he says. He speculates that the focus on text analytics stems from the fact that it’s the most prevalent form of communication captured as data at the moment. “Perhaps as we attach emoticons, images and video to the text we’ll get a more complete picture, but we have to start somewhere.”
Beth Schultz (@beth_schultz) says since Grimes’ take is that text is the largest source for harvesting sentiment, it stands to reason, then, that it’s often the starting point for a sentiment analysis program. “Or do we also see sentiment analysis initiatives launching with audio and video sources in mind, as well as images – typically, that is?” she asks.
And Ariella Brown (@AriellaBrown) says she likes to apply concrete examples to principles. So she wants to know if this post on social position with its map of Tweets would constitute sentiment analysis.
