Take Your Canary Down the Opinion Mine

Nowadays, social media and blogs have never had a stronger impact on products, businesses, and even individuals. Moving in this direction, it is becoming increasingly insufficient to only dig into quantitative data for insight. A growing trend is currently shifting towards sentiment analysis, accounting more for opinions and emotions.

I may not be entirely exaggerating when I say that something like Twitter is capable of making or breaking a brand, as people are now starting to refer to these social mediums for data. Had I read through Tweets about the movie Funny People before seeing it a few weeks ago, I probably would have made a different choice. These days, companies have very little choice but to dig into these sentiments and rethink their marketing and PR strategies to respond accordingly.

There is no doubt that today’s tools are able to provide us with a more complete framework that incorporates both quantitative and qualitative data. But the degree of accuracy and sophistication of these tools are questionable when it comes to their qualitative aspect. Most available sentiment analysis tools operate by 2 simple steps- scanning then categorization. Digging into open ended results, they are able to search for specific words then categorize them as either positive or negative. As simply put by Alex Wright of the New York Times, Love is good, Hate is bad.

With this data scanned and categorized, you can compare your positives and negatives to get a general picture of your brand perception. Unfortunately, people like me can really mess up your data. While these linguistic tools perform simple binary analysis, they fail to capture idiomatic aspects like humor, irony, and sarcasm. I recently Tweeted about the movie: “Funny People, not so funny”. Will this be categorized as a negative because of “not” or positive because of “so funny”? How about if I said that the new BlackBerry is “sick”? Can the tool detect the difference between that and saying that it makes me sick? This is where these automated tools can fall short.

At this point in time, we must accept that these tools will not provide us with the 100% accuracy that we seek. The only available tools with this sort of accuracy, in my opinion, are humans. For the mere purpose of positivity, let’s just say that a bit of accuracy is always better than none. So if we consider ourselves miners, let’s think of these tools as our canary birds. We can take them down the opinion mines with us, but let’s not put our full dependence on them just yet.