In early November you may have read the interview I did here with Seth Grimes on the Future of Sentiment Analysis. Seth provided some great insight into describing the current state of the sentiment analysis space, where it is headed and what role social media plays in the grand scheme of things. Soon after that post, I was able to attend the third annual Sentiment Analysis Symposium, founded by Seth, in San Francisco.
I understand the basics and importance of sentiment to marketing and data mining, but it was interesting to listen to many of the most passionate experts in this field discuss the nitty gritty of this growing area. Here are a few of my take aways from the event:
Sentiment analysis is extremely dependent on social media.
Just about every presentation, no matter what the topic, referenced social media as the primary fuel driving modern sentiment data. Every status update, tweet, blog post, product reviews, etc., … social media is providing data at an unprecedented level. Social media is oozing sentiment rich data. As the amount of data increases the patterns, trends and context that can be extracted from it will become more important for businesses of all sizes.
Twitter is an ocean of sentiment data.
Of all the social networks, Twitter was mentioned most frequently by speakers. The rushing river of 140 character or less posts provide the most conversational data on the web by far. A single tweet on its own may have some value when building a sentiment data set, but when you analyze the hundreds of thousands of tweets that are posted each day there is much more meat on the bone to chew on. Also consider that most of the information on Twitter is public in comparison to networks like Facebook which is primarily private. It all makes Twitter a very attractive place to gather and analyze sentiment.
There is much more to sentiment data than what we find in text.
Although sentiment analysis is usually associated with extracting information from the emotional context of text, the social web is much more than just text. Although NLP (Natural Language Processing) plays a huge role in the current technology used to gather and assess sentiment it is not enough. No matter how accurate software gets at understanding the emotional meaning in text on the web there are other elements that need to be factored in. How should you treat a “like” on Facebook? How about if the data you captured is a retweet rather than an original tweet? Can a sentiment score or value be given to a social action like this? Even more difficult are the potential complexities of trying to extract meaning from an image or a video. Imagine trying to understand the emotional context of a powerful photograph making rounds on Facebook. We only scratch the surface by understanding the textual data.
Sarcasm is sooooo cool!
No, not really. Understanding sarcasm continues to be a tough nut to crack, but analysts and technology providers are getting better at it. Understanding the context around key words or phrases is critical in accurately assessing the emotional value of the content. Context is equally important in being able to understand if something that seems to be positive is negative and vice versa. Take my heading for example “Sarcasm is sooooo cool!” At first blush it might be identified as saying something positive, but when the context within the rest of the content is read there is evidence that my intention is to express the opposite. This happens more often than one would think. Being able to effectively pull apart sentences and phrases piece by piece to identify the subject, the context and the sentiment is likely the direction many technology vendors are going to take. By my observation, some of them are already doing it now.
It all must tie back to business goals and existing processes.
Ultimately, as with social media marketing, you need to consider how your efforts are going to tie back to business goals and processes. Are you gathering the data to help you understand what people feel about your brand or are you leveraging it to help you better understand an emerging problem you may be able to solve with a new product?
There are a number of reasons you might consider gathering and analyzing sentiment data, but you need to think ahead so that the data you gather is inline with your desired outcome.
Also think about how this data might empower existing processes and departments. Is there value in connecting sentiment data to your existing CRM so you can see what previous experience a customer may have had with your brand and why he is unhappy with you? For instance, I saw an example of a company analyzing the transcript of a customer service phone call. The gentleman on the phone was clearly unhappy and because they knew who he was they were able to pull records of previous interactions he has had with them which pointed to why he was voicing his discontent. Does that make the customer service rep better at making the situation right? Likely.
In general, the tools, trends and best practices for sentiment analysis have a lot of room to grow. Social media will continue to play an important role in how quickly this space grows and ow we apply it to business. If you want to check out some of the sessions from the Sentiment Analysis Symposium you can find videos of most of the sessions here. For those of you who enjoy sarcasm, just imagine trying to analyze the conversation in the video below. Enjoy.
Have You Registered For Explore Dallas-Fort Worth?
Don’t miss a day of intensive learning with some of the leading thinkers and practitioners in the digital marketing and social media marketing space. Copyblogger’s Brian Clark, Edison Research’s Tom Webster, Edelman Digital’s Zena Weist and more headline one of the leading digital and social media marketing events of 2012, February 17 in Dallas, Texas! DON’T WAIT TO REGISTER! The first 100 to do so get an incredible discount! Reserve your seat today!
VIP Explorers Club
- How One Company is Using Random Acts of Kindness to Build Their Brand
- A Guide to Building the Best Possible Facebook Custom Audience with Jon Loomer
- How Marketing Will Change at the Incite Marketing Summit
- Book Review: ‘The CMO’s Periodic Table’ Reveals the True Elements of Marketing
- How to Use Reddit to Drive Traffic to Your Site