Slashdot Mirror
Algorithm Deduces Drunk Tweets From Geolocation, Behavioral Data (thestack.com)
An anonymous reader writes: Researchers have devised an algorithm for identifying people who post on Twitter while drunk at home, using both geolocation tags and behavioral data. The researchers analysed alcohol-related tweets from New York City and Rochester, and found that the tweeting drinkers in Monroe were likelier to be out of their houses than New Yorkers. The scientists concluded that the model could reveal important real-time information for public health research — creating a tool for improving a community's health, and using social media as a resource to spread positive health behavior.
Frist psot? Two drumk 2 tel;.
They seem to, but I'm always baffled when no one laughs.
Don't step on the baby.
Wisdom of the masses.
Efficient markets.
Social science.
Don't step on the baby.
Nabil Hossain is a Doctoral Candidate in Computer Science at Rochester. The experiment is concerned about a particular application of a more general Algorithm.
So, you ignorant twit, did you even read the Paper?
using social media as a resource to spread positive health behavior.
In other words, a resource for insurers to screw you for drinking "at home and/or at inappropriate hours", and for employers to give you a bollocking over the same.
If construction was anything like programming, an incorrectly fitted lock would bring down the entire building...
How are they identifying the drunks? Are they targeting people who tweet their ex at 4am?
Build a Man a Fire, and He'll Be Warm for a Day. Set a Man on Fire, and He'll Be Warm for the Rest of His Life.
Like i need another reason not to use Twitter. Sheesh
Have you ever fallen asleep at the keybhanusdiog?
this does not sound very scientific.
in the 1st place are these tweets from people who were really drunk? seems there was no real verification.
without verification of data used, how can one judge the accuracy of the "algorithm"?
even if there was verification, algorithm would only predict a probability, (eg. "algorithm indicate there is 80% probability that this tweeter is drunk") rather than a certainty.
given all that, it is rather premature to expect this "model could reveal important real-time information for public health research".
let alone "improving a community's health, and using social media as a resource to spread positive health behavior."
more real science, more honesty about limits of what can be known, and less exaggeration of claims, would be better.
btw who decides what is "positive health behavior" and the "community's" need to "spread" them? sounds rather orwellian to me.
It appears that RTFA is measurable. But the one thing that I see is an increase in DUI's and being used as a funds generator for certain Bad Actors.
Is Linux being used here?
Buy an intelligent thermostat they said, it'll save money and make you more comfortable they said.
Now they want to track when I'm anywhere "to help determine if I'm away."
Now they want to analyze my tweets and use my location data to see where I am?!?!?
Get the fuck out of my life. I do not want to be a statistic, I do not want to be a data point. Now get off my lawn.
Harrison's Postulate - "For every action there is an equal and opposite criticism"
It's amazing (and scary) the uses people are finding for U of R's research in AI. It is a little creepy being able to make fairly accurate guess as to one state of sobriety (or lack thereoff). Then again, people who are tweating at odd hours are already more likely to fit the profile of being in a drunken state (. What is truly scary is combined with date storage, the ability to profile people. Imagine HR purchasing access to a database of info like this to prescreen applicants. Insurance companies will have a field day with this stuff. Think I'll keep my twitter account silent and my life private. (Facebook anyone)
"Imagination is more important than knowledge" - Einstein
This was not actually checking to see if a person was drunk, it was speculation that a person was. Similarly, the Government won't really check to see if you are drunk. They will check to see if you are a threat in any way, and then _claim_ that you are drunk.
When "drunk" has no definition provided, and the Government claims that anything from .5 and higher is drunk (1 drink), it should be obvious that this really is not about anyone's health.
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
Deduction creates a conclusion that is necessarily true through logic, but in this case machine learning comes to a conclusion that is only probably true.
It's elementary, my dear Watson.Elementary, my dear Watson
Seems like a great service.
Don't waste your vote! Vote for whoever you want, unless you live in a swing state it won't matter anyways
When the hashtag #ThingsThatDontGoWellTogether was trending, Dave Chappelle tweeted, "Twitter and alcohol."
Then he left twitter not long after that.
Smart man, that Chappelle.
"Who are you?" "No one of consequence." "I must know." "Get used to disappointment."
I want to write an algorithm that uses data from wearable fitness bands with heart rate monitors to identify people who masturbate at home. Maybe the University of Rochester will fund my research.