Info 202 Fall 2011

The recent surge in the use of social networks has produced an increasing torrent of web content. Such content has remained untapped for sometime, until recently. Users tend to post comments and notes the same way they would speak in the physical world,  which may have been the primary reason why harvesting such information has been a challenge. 

Naturally, it didn't take long for technology to evolve, producing a number of techniques such as NLP and text analysis to overcome this. This article talks about the results of using such techniques to analyze brand sentiments from web conversations about the top 10 technology companies. The analyzed content included Tweets and Facebook posts and was done over a 12 month period. Such information can be extremely valuable for companies wishing to have a broader insight to what their consumers want, offering a chance to scale much larger than any feedback survey ever could. 

A number of smaller and more experimental websites offer a similar functionality. Individuals who are curious about peoples general reaction to something can use services such as TweetSentiments.com to look up the general tone in relation to a topic, person or event. 

On a larger scale, there have been some successful attempts to apply sentiment mining to information on world events. A recent BBC article indicated we may now have the ability to even predict the likelihood that events would occur based on the analysis of news outlet publications. Such prediction technology is far more complex than the last two examples, which is why its uses will not likely be widely commercialized in the near future. Nevertheless, the accelerating development in technology products can bring to mind some interesting applications for commercial and personal use. In  a time when we 'Tweet' with each other, 'talk' to our iPhone assistants, and 'email' the person sitting next to us, is it possible that, in just a few years, we will be running complex algorithms that will predict likely actions of those around us? The probability of a new product success, or the satisfaction an audience with your work, or even the progress of an election campaign. Will computational processes exceed our human ability to infer sentiments? Perhaps they will, the possibilities are endless.