Meeting minutes
<koji__> quit
[slide description]
osamu: slide 5 [project and related works on data architecture]
… the back ground of this talk
… we need your voice on this presentation.
takuya: after the talk on my presentation, please give us your opinion.
… [self introduction]
… [slides description]
… slide 4: issue to confirm the credebility
… slide 5: trust in business
… building the trustworthy
… slide 6: our approach is checking by physical world
… slide 7: architecture with three endorsement layers
<Geun_Hyung> please share the link of present material..
takuya: slide 8: Endorsement graph/flow of each data
… slide 9: endorsement graph generation, with each attribution data
… slide 10: presentation description
… slide 11: conclusions
… slide 12: discussion points, use cases,stakeholders, ecosystem
gendier: trust we handle many years. two main concers
… trust signal individual involved. makes difficult problem
… 2nd is technology point, TXT score of trust. big problem is bad score coming
… do you have any solution?
takuya: trust list is not easy to make
… so usually social media, people will not expose much data.
jonathan: use case..
brentz: trust over internet has similar problem, do you know?
takuya: our main characteristic is endorsement layer.
… ACDC will help
<brentz> Trust over ip: https://
annette: our community group also discussing on trusted TXT assigning score.
… please check web site entity. little bit issue with scoring people.
… endorser should has reliable source
takuya: this solution is adding endorsemnet, and define data model etc.
bmay: can I trust the integrity? next level can I trust of orgin of signal.
… easy path of signal, more trustworthy path. question is trust and cofidentiality
… believe of privacy, graph of endorsement, becoming social issue.
<gendler> +1 to bmay statement on privacy
takuya: this solution is signed data. it is not easy real trustness.
osamu: our proposal model is current internet.
… mashup the verifiable data is our approach. more trustable will recognize with plural data
shigeya: I am co-chair. we want have ultimate solution. but at least, verifiable data, distiguished data.
… we can one step up by our approach.
dhuigens: according to your presentation, we assume the web App run correctly.
… security model now . VC at web app is not good. do you agree?
osamu: of course, we need some end-point trust. which rely on other, we focus
… it si now out of course, requirement is our scope.
<Zakim> Bert, you wanted to ask about levels of endorsements: can I endorse something partially, e.g., to say I believe it is 50% true, or that, if an article claims something is A out of A, B and C, I know that at least it is not C?
Bert: partial and negative endorsement. ex endorse 80%. could realized?
takuya: data model is very difficult.
… this is not easy problem. but need to do it.
osamu: evalution metric is always problem. on the contrary, how we can design is our part.
<Zakim> gendler, you wanted to clarify my comments on trust.txt
gendler: base system is the relationship between two. which base on trust TXT.
takuya; what is TXT?
gendler: JSON document, technicaly machine readable domain
osamu: I need originator profil led by Keio univ on advertisement. Trust of two party is described there.
<annette_g_> To clarify the notes about my comments: I pointed out that trust.txt does not issue scores. It simply enables entities to state relationships with other entities. I also mentioned that the proposed system has two issues; first it does rely on endorsers being trustworthy (which is different from simply identifying itself), and second it requires identification of relevant endorsers for a given message, and it's not clear to me how[CUT]
<shigeya> Originator Profile Project: https://
ken: general comment. need more use cases. it will easy to understand. for example fake news.
… to help to concrete idea such as model. generic study only is not good approach.
<gendler> Annette; I'm sorry I didn't properly address your latter points! I would agree with them entirely. They try to offload this to the economic incentives of trade associations to curate their members properly for trust, but that was overly optimistic in my estimation as well!
takuya: we will select harmful example for W3C. scoring is useful for viewer.
takuya: thank you for your help on discussion.
@@: use case my sense of high/low, people's understanding is very different each. this is not good for trustworthy
takuya: incentive is required to check their honesty..
kojima: should we connect/divide both opinion?
@@@: this is not technical problem.
osamu: please introduction of trusted-web
shigeya: our trying on trusted web is extend the trusted data. if we extend our trusted data, it will ehlp gap.
… we need effort to fill this gap. we found organization has various document in terms of verifiable. Area of trust is much mor easy for everyone.
… this is a multi-year project. now we issue Japanese version in August. English version will come soon.
[adjourned]