23 comments

[ 3.1 ms ] story [ 72.1 ms ] thread
A sarcasm detector. That's a real useful invention.
(comment deleted)
Yes, if we ever want our digital assistants to understand language.
Joking aside, TTS engines could benefit from that.

Even though I can see how a false positive would make a great joke out of a serious text.

This is the perfect response. I assume you're being sarcastic, but I can't tell for certain. Even most humans have trouble detecting sarcasm in print - would a sarcasm detector?

Since sarcasm is the biggest weakness of sentiment analysis, I don't have much faith that sentiment analysis tools will produce truthful results when aimed at social media.

"I assume you're being sarcastic, but I can't tell for certain"

What if the person you're responding to (for the sake of argument) really doesn't know?

I mean, it can take some time to find out if an invention is truly useful.

So if we point it at that statement and demand a verdict, it's sort of an undecidable problem.

This is the perfect response. I assume you're being sarcastic, but I can't tell for certain.

No, it is simply a good response. I'm British (and have special skills) and can inform you that satherx is either rather obviously sarcastic or a lemon.

Sadly there is not enough “conversation context” in your comment to use as a test in this analyzer.
"Internet Argument Corpus (IAC) is a publicly available corpus of online forum conversations on a range of social and political topics, from gun control debates, marijuana legalization, climate change, evolution, to name a few (Walker et al. 2012). The corpus comes with annotations of different types of social language categories such as agreement/disagreement (between a pair of online posts), nastiness, and sarcasm. There are different version of IAC and we use a specific subset of IAC in this research. Oraby et al. (2016) have introduced Sarcasm Corpus V2, a subset of the Internet Argument Corpus V2, which contain 9,400 posts labeled as sarcastic or non-sarcastic (balanced dataset)."
I wonder if there's a corpus labelling comments and posts based on the context of, "/s". Because labelling a comment as sarcastic is a great indicator the poster understands how sarcasm works.
good job.. what, with the lack of context and all.
Finally, visitors from Betelgeuse no longer have to rely on a human interpreter.
The computers can save us from Poe's Law. ;)
"Without a winking smiley or other blatant display of humor, it is utterly impossible to parody a Creationist in such a way that someone won't mistake for the genuine article." - Nathan Poe
Who is going to write the app? It is sorely needed.
(comment deleted)
Please use it to analyze this thread :)
sucking all the fun out of comment sections.