Steve writes:

"I'd be interested in more detail on this concept as you reference it.

In an early form, one might:

1) Take fact checking websites like 
WaPost's<https://www.washingtonpost.com/news/fact-checker/wp/2016/09/21/the-2016-election-fact-checker/>
 and encode the claims in a language like 
CycL<https://en.wikipedia.org/wiki/CycL>.   This would be a human curated 
database, tuned by ontologists and journalists.   It would be exposed through a 
client library that anyone could use.

2)  A company like Twitter would make call to this library to invoke a natural 
language parser to turn tweets (near raw text of not quite English) into 
propositions in first order logic.

3) Before rendering text as XML/HTML, they would invoke a logic engine via a 
library call to satisfy the propositions from #2 using the database in #1.  The 
library could add attributes to the XML and some simple CSS  rules could 
control how it was shown in the users' web browser.

The Washington Post talks about shading of facts, selective telling of the the 
truth, and use of legalistic language as determinants in rating truthiness.   
One way to do this would be to have #1 built into typically-observed sets of 
clauses -- frequently observed claims that people repeat without significant 
embellishment.  "Hillary says the working white are deplorable" would be a low 
truthiness tweet whereas "Hillary says some Trump voters are racist or 
misogynist  and others are frustrated because nobody cares about them" would be 
a higher truthiness because the latter captures more of the original 
text<http://www.latimes.com/nation/politics/trailguide/la-na-trailguide-updates-transcript-clinton-s-full-remarks-as-1473549076-htmlstory.html>.
   The truthiness could be communicated by suppressing text when it was >= 3 
Pinocchios, forcing the user to click on an icon to reveal the original 
nonsense.   Of course, the natural language processing would have to be pretty 
good at normalizing propositions into a form consistent with the database, and 
the database would need to be up-to-date with the candidates' various public 
statements.

Marcus


============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

Reply via email to