@jasper Well i guess it depends on how you want to visualize it. The point of the top bar is that the right in America is extremist and crazy and the left in America is extremist and crazy and there is little in between.
Thats why both halves are bright blue and bright red rather than a shade in between. To show the majority are extremists on both sides.
But yes I agree both sides in America are almost entirely comprised of extremists, certainly a majority. Trump is a prime example of that.
@jasper Yea, they did, abut 50% of them. Which I'd say is reflected by the top half of the image almost perfectly. :)
@Gargron Yea I guess if you want to avoid any training period then those wouldnt be good choices. But still there are unsupervised classifiers you could pick.
Stemming is language specific but if you use a library that shouldn't be a problem, presuming you can detect the language reliably. If not you can fall back to not stemming i guess. But even with your current approach I suspect you'd see an improvement with stemming.
Personally what I'd suggest is a combination of supervised and unsupervised. Unsupervised being a first-pass, then as users block certain messages over time the supervised learning algorithm would improve and make suggestions.
Anyway thats my 2 cents.
@Gargron I guess that depends on just how much effort you want to put into it. Naive Fisher Classifier or
Bayes classifier after running through a Stemmer would be the simplest that comes to mind while still being very effective.
@cwebber Story of my life right there.
@Gargron Any reason you decided to go with this algorithm rather than one of the traditional algorithms used to identify spam and/or similar messages?
I suspect youll find a lot of false positives with your choice of algorithm.
@peterdrake Incorperate actual rpg elements. Complex stats, treasure, crafting, character customization. All the elements a game needs rather than relying purely on a single gimmick.
@snder Bedankt
@lebronjames75 I can eat plain white rice sothered in sauce sauce with nothing else on it and I'd stil have a smile on my face.
I like the idea and would support it. My thoughts.
* False-positives need a good way to handle via UI
*User should be able to turn the feature off
* You're algorithmic approach could probably be improved on. The current approach could be easily circumventer I suspect unlike more traditional approaches like Naive Fisher Classifiers and such.
* By only storing the last 10 messages its also easier to circumvent than a larger value. Perhaps make this configurable?
@Gargron Cool ill check it out and give some proper feedback.
@Gargron First I heard of the idea. If its captcha based + Bayes filer to auto-flag I'd support.
@Expat1975 Already on my way :)
@Expat1975 I <3 Tsing Tao. Havent seen one of those here in the Netherlands yet though :(
Jeffrey Phillips Freeman
Innovator & Entrepreneur in Machine Learning, Evolutionary Computing & Big Data. Avid SCUBA diver, Open-source developer, HAM radio operator, astrophotographer, and anything nerdy.
Born and raised in Philadelphia, PA, USA, currently living in Utrecht, Netherlands, USA, and Thailand. Was also living in Israel, but left.
Pronouns: Sir / Mister
(Above pronouns are not intended to mock, i will respect any persons pronouns and only wish pronouns to show respect be used with me as well. These are called neopronouns, see an example of the word "frog" used as a neopronoun here: http://tinyurl.com/44hhej89 )
A proud member of the Penobscot Native American tribe, as well as a Mayflower passenger descendant. I sometimes post about my genealogical history.
My stance on various issues:
Education: Free to PhD, tax paid
Abortion: Protected, tax paid, limited time-frame
Welfare: Yes, no one should starve
UBI: No, use welfare
Racism: is real
Guns: Shall not be infringed
LGBT+/minorities: Support
Pronouns: Will respect
Trump: Moron, evil
Biden: Senile, racist
Police: ACAB
Drugs: Fully legal, no prescriptions needed
GPG/PGP Fingerprint: 8B23 64CD 2403 6DCB 7531 01D0 052D DA8E 0506 CBCE