Can Sentiment Analysis Be Effectively Automated?

Listening to the voice of the customer is a major initiative for many companies. Sentiment analysis is a growing component of the tools to assist in this effort. Sentiment analysis is defined by the Wikipedia as “the application of natural language processing, computational linguistics, and text analytics to identify and extract subjective information in source materials.” It is getting a lot of hype and there are conferences and events springing up to cover this buzz. Brand awareness is one of the major use cases.

However, sentiment analysis is also attracting its doubters. Peter Knoblock writes in the Smart Data Collective, Sentiment – A Potential Mirage on the Data Horizon. He writes that social media is relatively new and, as often happens with new technologies, people tend to implement old methodologies when trying to interpret this new type of information. He continues, “a common mistake is people try to turn social media data into quantitative information. The most common trend is focusing primarily on what is positive and what is negative.”

Automated Sentiment Analysis is Only a Piece of the Brand Awareness Puzzle

Peter notes that while positive and negative sentiment can provide a form of brand awareness, it is only a small piece of the puzzle. And this assumes that the sentiment was properly understood by the machine. We will come back to ask this question again.  But assuming the machine got it right, Peter asks some useful questions including: what is driving positive comments and what is driving negative comments? What are the themes surrounding the conversation? What types of terms keep showing up in conversation? What types of events or statements drove an increase in activity? He adds that looking at the themes and terms within your data can give you a richer brand awareness experience.

Can a Machine Really Understand Sentiment and Its Relevance to Brand Awareness?

Machines can be taught to understand words but what about context? There ha sbeen some progress in this space but more likley needs to be done. Someone could be happy your brand is tanking. They could be sad your competitor is having trouble. They could be angry that people are angry with your brand or they could be pleased with this turn of events. Then there is sarcasm or is that was I was just doing?  

I have heard that sentiment analysis tools take a stand only on a minority of content they scan. Often return some form of “not sure” as the result. To be fair, it is still in a primitive stage. Most vendors will acknowledge that there is a ways to go. Machines are good at going through masses of content to uncover things for people to look at. People are still best at determining meaning.

I have to admit that I have not used a sentiment analysis tool so I could be missing something. What has been your experience with sentiment analysis? 

 

Comments

[abdulla99]

http://www.imassagetherapist.org/directory/index.php?prov=Quebec