Sentiment Analysis

By Aaron W. Rockwell, Staff Writer

Since children, we are inundated with the concept of Yin and Yang, Good vs Evil, the Ninja Turtles vs Krang (& Shredder). Only as an adult did I come across the concept of sentiment analysis, and only recently have I started daydreaming about the concept instead of listening to lectures.

Sentiment Analysis: the process of computationally identifying and categorizing opinions expressed in a piece of text, especially in order to determine whether the writer’s attitude towards a particular topic, product, etc., is positive, negative, or neutral.

The Hedonometer is one such sentiment analysis that created its word bank by paying folks to rank 10,000 words on a  scale between one and nine (assumably odd because then five gets to be Switzerland). Each word had at least fifty people rank it.

Laughter ranked as the happiest word and Terrorist ranked as the saddest word.

SentiWordNet: this sentiment analysis uses computer modeling to start with a couple hundred words and then figure out the sentiment value of other words based on context. This is actually a free downloadable resource.

Best yet, here is a sentiment analysis tool where you can type in sentences, and it will come out with whether the sentence is positive or negative. This was created based off of movie reviews!

Overall, these methods are cool, but obviously, single words lack the ability to understand context, e.g., sarcasm or reverse qualifiers. As computers get smarter and as larger populations create more curious humans, I foresee the future as having words that have more quantifiable attachments, and I am pleased by this.

 Ideas and Concepts Currently Unexplored:

-Do individual letters have sentiment?

-Sentiment Analysis on Yearly Reports to see if positivity/negativity correlate to future share price.

-Sentiment Analysis on individual book sales to see if positivity/negativity is a factor.

-Reviews that are based off of review sentiment scores instead of that dinosaur five-star system.

If you were wondering: this article came up as 60% positive.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.