For years, sentiment has been a widely used measure of how customers view a company’s products and services. But sentiment analysis has inherent flaws. First is what it cannot tell you because it only considers a small amount of the available data. Only about 25 percent of posts actually contain sentiment, either positive or negative, which means three out of four posts are neutral, revealing no sentiment, and are effectively being ignored by the analysis. Thus, decisions are being based on what only a quarter of the posts are saying.
Another problem with sentiment is statistical confidence in the data. Simply stated, all methods of sentiment analysis rely on example data that, whittled down, reveals a low level of confidence about the sentiment being identified, either positive or negative. Data with such low confidence is a poor foundation for sentiment analysis.
Bottom line — sentiment analysis is not inherently bad. In fact, for particular types of questions, it may be the right approach. But if you use it, you need to be sure that it’s the right tool for the job, that valuable data is not being ignored, and that the method of sentiment analysis is built on sound data.
A much more statistically reliable approach is semantic analysis — a way to distill and create structure around mountains of unstructured data, such as blog posts, social network chatter, tweets and more, without preconceived ideas of whether or how they are related.
Semantic analysis allows you to cluster different data elements based on similarity, rather than preset classifications such as positive, negative and neutral. This helps you uncover important information like what exactly people are saying about your product or service; where and how they use it; and enhancements or new offerings they’re interested in. This type of valuable information can drive product development, new revenue streams and strategies for marketing, advertising and media planning.
Click here to read an interesting report that digs deeper and compares Semantic vs Sentiment Analysis.
No related posts.
Get email updates
Get RSS updates
Get podcast updates
Plus me
Friend me
Follow me
Network with me
Paul, in my opinion, any decent sentiment analysis applies semantic analysis. Further, in my opinion, the statement “Simply stated, all methods of sentiment analysis rely on example data” is simply wrong. Many sentiment-analysis methods rely on linguistic artifacts — lexicons of words that indicate subjectivity or sentiment and syntax patterns that link sentiment to subject — in addition to providing scores that aggregate measured sentiment. Better methods will allow arbitrary sentiment classification, not only into positive/negative/neutral tone categories but also into emotion categories (e.g., angry/happy/sad) and intent indicator categories (e.g., plans to renew service/plans to cancel/upgrade candidate).
There’s no opposition between the two categories, sentiment analysis and semantics analysis.
Seth, http://twitter.com/sethgrimes
Seth
First off thanks so much for commenting on my blog – thats huge for me!
here are some thoughts back regarding your comments
regarding this quote “”Simply stated, all methods of sentiment analysis rely on example data” is simply wrong. Many sentiment-analysis methods rely on linguistic artifacts”
Yes, but how do you know if those methods work or not? The only way to know is to use example data to test your system. This is precisely why the paper states “All methods of sentiment analysis rely on example data to design, TEST OR VALIDATE the analysis.” Without example data, you are just guessing at a solution to some unknown problem. And, as soon as you use example data, you run into the statistical confidence problem detailed in the paper.
“in my opinion, any decent sentiment analysis applies semantic analysis”
I don’t disagree. In fact, I would go a step further (as did the paper) and say that sentiment analysis IS semantic analysis. It’s just a form of semantic analysis that has a very narrow view of meaning and relies on very noisy data (as measured by statistical confidence).
and regarding …
“Sentiment analysis is not inherently bad; for particular types of questions, it may be the right tool. But if you use it, make sure the data underlying the analysis is sound and valuable data is not being ignored.”
When sentiment analysis is the right tool for the job, we use it. The important thing is to understand WHEN it is the right tool for the job, which involves understanding its problems, and to understand what the alternatives are.
thanks again and let me know if you want to speak live!
p
Paul, I’d say the word “example” threw me off, but so tell me, how do you “test or validate” semantic analysis other than with a) some form of “gold standard” (example) data, whether machine or human annotated, or b) inter-annotator comparisons, again whether you’re comparing to human or machine annotation that, in this second instance, is not considered “gold standard” definitive?
I can’t think of any way than my (a) and (b), and if you can’t put forward of any other way, then any complaint about the use of “example” data to test or validate sentiment analyses would equally apply to semantic analyses.
[...] such as cluster analysis, a highly complex way of assessing the gathering of words and things, and semantic analysis, which is a form of sentiment analysis but with a wider Big Data lens. The Big Data impact [...]