« Dysfunctional Communication Leads to Dysfunctional Work Places | Main | Wish you were here »

Try sentiment analysis

This week’s post on dysfunctional communication has been on my home computer for some time. Was it too negative for a civility blog? How about rewriting…flipping it 180 degrees? It wouldn’t have the same impact.

It took me two weeks to link in my mind that post and a ‘sentiment analyzer’ a fellow Iowa State communications specialist had provided. This week I plugged the text into the analyzer. It came up ‘Overall Sentiment: Positive’. Enough of a green light for me.

Sentiment analysis is
a statistical analysis of the adjectives you use, tone, cues, polarity, culture and more…gauging the attitude or opinion of the communicator. (This is what I glean from a document by people from Google Inc. and the Department of Computer Science at Stony Brook University. It’s noted at end of this post.)

Shel Holtz, Accredited Business Communicator, principal of Holtz Communication + Technology wrote on his blog, (http://blog.holtz.com/index.php/update)---
Sentiment Analyzer
Enter the text of an article and watch the Sentiment Analyzer parse the document and then let you know whether it was mostly positive or negative, highlighting the elements that fall into each category. Notes David Phillips, “This kind of development is useful for analyzing sentiment of news articles, blogs and other content, which is its primary purpose but it also has applications in evaluating style and bias all of which are very useful to the public relations industry, regulators and watchers of political sentiment on and off line.”
http://netreputation.co.uk/sentiment/

Give it a try
Plug in an email or any text. It gives you positives and negatives. Remember it’s a computer…algorithm and all that. But it might be better than your own biased opinion.


For the people more technical than I

Large-Scale Sentiment Analysis for News and Blogs
Open-source RapidMiner for data mining and analysis

Comments

I like your approach using content analysis. I think setiment is another work for values. I used a similar approach when developing the workplace attitudes test. It was developed by interviewing jerks at length. This included job-hoppers, those with low job ratings, embezzlers and even a few that brought lawsuits. I was looking for shared values or sentiments. Perhaps you would enjoy my blog that addresses this at www.workplaceattitudes.blogspot.com