credweb

Minutes: Credible Web CG SUBGROUP 1 on Inspection (16 July 2018)

Topics

  1. Outbound references

Resolutions

Minutes

# [Sandro Hawke] Agenda https://credweb.org/agenda/20180716.html

# [An Xiao Mina] https://credweb.org/cciv/#outbound-references

# [Sandro Hawke] recording to cloud, all agreed

Outbound references

# .. footnotes, endnotes, links

# .. reputation of the things you're linking to

# .. credco started with indicators for health & climate -- science topics

# .. in breaking news this might be completely different or not apply

# .. esp when tweet is deleted, then what to trust

# .. the ones we tests in CredCo

# .. we looked at the articles cited, how Representative are they of the science article being references

# .. are you quoting an outside expert in the field? a best practice in journalism

# .. looking at Impact Factor of journal linked-to

# .. IF is very roughly measure of value/influence in the field

# .. Sentiment

# .. So yes, Tone of Citation would be good

# .. but we need to look at when there is disagreement in the field

# .. gets back to question about relational indicators

# .. and emotion indicators

# .. tying that together would be good

# [Sandro Hawke] now going one at a time.

# .. 1 Source Types

# .. could be built out more for more nuance

# .. per citation -- we asked them to look at 3

# .. didnt correlate

# .. not in WebConf paper

# .. not sophisitcated yet

# .. Like good practice in science reporting, to quote people from outside

# .. Obviously quotes could be fabricated

# .. ideally the conversation happens EG on Twitter

# .. you want to link to the quote In Context and Endorsed by quoted person

# .. ideally

# .. I think it was a group in Canada

# .. Number of quoted sources, Number of links

# .. just seeing if sheer number is interesting

# .. like Number of ads.

# .. I don't recall the story here

# .. 9.7 Contains original quotes

# .. Not in WebConf, although it seemed promising.

# .. not from the study

# .. related/overlapping with 9.2 and 9.3

# .. not necessarily a quote

# .. like a statement they refused to comment :-)

# .. 9.8 Contains Video Embeds

# .. 9.10 Contains Image Macros

# .. 9.11 Contains Attributed images

# .. 9.12 Contain Original Images

# .. all about embedding visual content

# .. intuition is that these will be positively correlated, but we didn't study this yet

# .. might be different if embed is extract from study, or something else

# .. 9.9 Contains Link to Scientific Journals

# .. we did study this

# .. (not sure offhand the result)

# .. but related to 9.14 Represtation, which we did

# .. 9.13 Agencies for Authority

# .. not sure who the "I" is, probably folks into 9.1 source types

# .. eg CDC is authoritatinve agency

# .. related to eg IFCN inbound links

# .. Idea is - you have reputable agencies that are not eg IFCN members

# .. contains (embedded) content

# .. 9.14 Accuracy of representation of source article

# .. more in detail

# Sandro Hawke says: : Citations - this whole topic - in general, negative references are different. But mostly we're talking about borrowing authority. YOu want to say something credible so you point to someone else saying the same thing.

# .. all of these seem to have some reputation element -- your reputation depends a bit on the reputation of who you cite

# .. 9.15 Academic Journal Impact Factor

# .. this was an explicit reputation representation

# .. Mike Caulfield has his students do something like this, checking on Google Scholar

# .. I.F. isn't perfect, but something like this seems machine-doable

# .. that works for your outside metrics

# [Sandro Hawke] https://scholar.google.com/citations?user=3iUAjSsAAAAJ

# [An Xiao Mina] https://scholar.google.com/citations?user=z2UUZSkAAAAJ&hl=en

# .. so calling someone an expert in the field without an h-index at least 10 would be silly

# .. you can look at the top 1% by hIndex in each field

# .. but cancel next week given F2F