According to a March 2008 article in Information Today, a company called Magpie is gaining traction among big business with software they developed called Brandwatch. They a sentiment analysis, identify what people are saying about a brand or topic on the World Wide Web. For about $300, clients can sign on for a one month trial. Companies can capture a glimpse of the public perception within one week. The challenge, according to CEO Greg Palmer, is that most will be unsure of what to do with this information.
After reading this first article, I was both excited and overwhelmed by the possibilities available to someone interested in gauging perception. On the one hand, here is a tool that can serve as a virtual clipping service. However, instead of relying solely on print media, the conversations, thoughts, sentiment of anyone participating in the culture of the World Wide Web is up for grabs. Wow! On the other hand, I have a difficult time trying to identify the most relevant information that I pick up from a simple feedback survey. The idea of logging, measuring, and analyzing—figuring out what to do with the information that Brandwatch delivers– is a daunting task to say the least.
My overwhelm and excitement, I realized after reading two research articles that explore Sentiment Analysis & Meaning, is just the tip of the iceberg.
The article, “Taking sides: user classification for informal online political discourse,” by Robert Malouf and Tony Mullen, evaluates language processing techniques within online discussions. The article raised the challenge of analyzing the audience and presence as it relates to the natural language processing. Content in chat rooms, on discussion boards, forums, and social networking sites is “intended for its immediate participants in the online discussion at the time the discussion was produced.”
If a tool such as Brandwatch picks up a piece of a conversation from a social networking forum, there is great danger that the intention can be misinterpreted.Further, as authors Yoshikiyo Kato, Sadao Kurohashi, and Kentaro Unui point out in their article, “Classifying information sender of web documents,” these conversations are often anonymous or it may be very difficult to actually assess the credibility of the contributor. The purpose of the Kato et al. research was to develop a method that will classify information sender of web documents in order to assist with credibility analysis.
In short, Magpie is doing much more that building its brand as it helps companies build theirs. There is much research to be done on the topic of sentiment analysis on the World Wide Web. There are limitations to the research that does exist and further refinement is needed before Magpie might achieve its goal to “provide customers with information that is useful.” First, CEO Giles Palmer must make sure that the information is actually reliable.
McClure, Marji. “Magpie Builds Its Brand as It Helps Companies Build Theirs.”Information Today. March 2008: 21.
Malouf, R. & Mullen, T. “Taking sides: user classification for informal online political discourse.” Internet Research. 18.2 (2008): 177-190.
Malouf, R. & Mullen, T. “Taking sides: user classification for informal online political discourse.” Internet Research. 18.2 (2008): 178.
Kato, Y., Kurohashi, S. & Inui, K. “Classifying information sender of web documents.” Internet Research. 18.2 (2008): 191-203.
McClure, Marji. “Magpie Builds Its Brand as It Helps Companies Build Theirs.”Information Today March 2008: 21.