TPAC Breakout on Credibility
- Summary of Action Items
- Summary of Resolutions
<sandro> (not scribed)
Credibility Coalition Into
An_Xiao_Mina: Meedan, working on fact-check tools
... common framework for markup of source credibility
... effort to develop standard for 3d pty markup
... id indicators and use cases; est cooperation; generate and evaluate testing data
... misinfo is bad info shared without intnt to deceive (distingfuished from disinfo or malinfo)
... [slide showing current group members]
... looking at indicators, not data interpretation or assessment
... .claim review; draft indicator framework
... includes structure, content
ClaimReview in schema.org
DanBri: Google, schema.org
... started discussion last year re providing structured data for fact-checkgin
... schema.org, looked at fact-checking as a kind of reviewing. "claims"
... added a property on claimreview
<fab_gandon> any link to the slides presented?
DanBri showing the exampleGoogle search "hillary deleted emails" with annotated result
scribe: what's the canonical source for a claim
... trust project,
@@: where does the data live?
DanBri: in the fact-checker's site
... e.g. politifact
sandro: I've been involved in this work with my MIT hat, and more recently, thinking about whether there's a place for it at W3C
... we have a W3C community group, a meeting Friday morning
RobSanderson: there could be many sources of claims
danbri: hoping for the development of a fact-check commons
... there's a network of reviewers checking one another's work
ed: we're developing a schema anyone can write to
... creating a lens by which we can look at info
Ivan: you can have ratings of ratings orgs
ed: source, claim, and article credibility
andrewbetts: currently, claim review sits on its own; can we use linked data
... to annotate URL
danbri: claim review markup could have a pile of URLs, or wikidata could collect them
fab_gandon: any background in argumentation theory?
jenny_lee: a whole range of techniques
<reto> Create a schema:ClaimReview for a tweet: https://twee.fi/
manu: is there a roadmap?
<fab_gandon> I would volunteer Serena Villata : http://www-sop.inria.fr/members/Serena.Villata/
sandro: we have a very draft schema
<fab_gandon> ... for argumentation theory aspects
Richard_Varn: would this be used to enable search?
... often, source of the claim is a part of the argument. Is the original source part of structure?
aviv: we're interested to hear if your org is providing or interested in this info
RichardVarn: lots of data in English language grading for common core
... grading for whether something is a valid argument
sandro: a test before I can post to social media?
danbri: there's an arms race aspect
... with a listing of credibility indicators, there's a rush to acquire them
... think about how we can resist, e.g. giving scientists tools to avoid being cited by junk
Brendan: IAB techlab, in ads context, problem of newcomers being given trust equivalent to established actors
... registry of persisten identifiers
jenny: we're not the black box; we're offering framework by which others can do analysis
dsinger: think about ways this could be subtly poisoned
ben: a whole chain of claims; who verifies the authority of the source, identity, etc.
... and whither satire?
AnXiao: big question
<gkellogg> Relevant presentation on Post-Truth community response from DC 2017: http://dcevents.dublincore.org/IntConf/dc-2017/paper/view/525/626
danbri: we just added a satire tag to schema.org
jenny: Donald Trump's speech for Black History Month...
Richard: Some projects using sentiment analysis, NLP
<newton> FNC (http://www.fakenewschallenge.org/) has some labeled datasets that may be useful for training ML algorithms
Joerg: are you thinking about things changing over time?
sandro: not yet there
[End of minutes]