Time and again I stumble upon technologies that, even though very noble in their goals, are also used for malicious purposes - and with much greater success. Even though I haven't yet found any troubling uses for the technology outlined below, the potential for misuse - especially in authoritarian states - is definitely huge. So here is an initiative co-sponsored by a project at the University of Massachusets at Amherst:
We are compiling a list of oaths, dirty words, racist and sexual derogatory and other offensive terms to build new language models and software tools. The notable increase in public and private threats communicated via the Internet to Congress and the White House makes it important to do basic research that identifies the precursors of aggressive or violent behavior. Adding your uniquely offensive contribution to the list, with an associated rating from 1-7, with 1 representing a mildly offensive term and 7 representing the most foul term, will advance this work.
The site then lets you input a word/phrase and rate it on a scale from 1 to 7, 1 being the least offensive. In essence, this is crowdsourcing the gathering of foul/offensive terms so that computers can better detect them. This is probably done for worthy causes: Public Comment Analysis Toolkit [pdf], another tool from the same project at UMA aims at "enabling government officials to listen to and engage with the American public about regulations that impact their lives and businesses" (via regulations.gov).
I can only imagine how useful such technology could be for identifying the next generation of dissidents in China or Iran. It's a funny world we live in: every tweet is potentially a "precursor of aggressive behavior"...
Read More
(1)
HIDE COMMENTS LOGIN OR REGISTER REPORT ABUSE