A Sarcasm Algorithm That Detects Snotty Reviews
May 20, 2010 - By Jane McEntegart
" Sarcasm algorithm reveals the most sarcastically reviewed products on Amazon
They say sarcasm is the lowest form of wit and whether or not you believe that, the fact is there's a lot of sarcasm on the Internet. Sometimes, it can hard to tell when someone is actually trying to be serious. This presents more of a problem in some areas than others, like when reading product reviews. Sure, you could just look at the rest of the reviews, which would give that one glowing assessment a bit of context, but why do all that work?
Geekosystem reports that a team of scientists at The Hebrew University's Institute of Computer Science in Jerusalem has come up with an algorithm they say can detect online sarcasm. Dubbed SASI (semi-supervised sarcasm identification algorithm), the algorithm works by taking a handful of sentences that have been tagged as sarcastic and employs machine learning to make guesses as to other sarcastic sentences.
When tested on Amazon and Twitter, Geekosystem reports that SASI scored pretty well. The algorithm achieved a precision of 77 percent when processing Amazon reviews and Tweets.
The study found that the most sarcastically reviewed items on Amazon are Shure and Sony noise cancellation earphones, Dan Brown’s Da Vinci Code, and the Kindle. The researchers said that, though there are no rules for sarcasm, indicators are excessive use of capital letters and exclamation marks. "
Source: http://www.tomsguide.com/us/ebay-kin...news-6851.html
Bookmarks