Re: AI KR meetings 2021

Thank you LG

Glad you think it could be useful, and no, we have not made a start as a CG
on that yet but I threw it as a possible
agenda for the next call

I am following up on a meeting organized by the Public Standards in the UK,
and if someone else in interested, we
could make a contribution via email as a CG


  Artificial Intelligence and Public Standards  Committee of the UK Gov
Feb 2020

https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/868284/Web_Version_AI_and_Public_Standards.PDF
The document here actually calls it Register
There are 10 instances of Register in this document, some examples below

  This represents a missed opportunity. The marketplace could offer a range
of tools to help providers assess if AI products will support or undermine
public standards. Canada, for example, operates a register of responsible
AI companies.72 The marketplace could also allow AI products and services
to be classified according to certain features, such as explainability.
Such tools would help public bodies navigate the range of products and
services offered. In discussions with the Committee,
 Crown Commercial Service (CCS) officials expressed a desire for the
marketplace to

  In its report on Algorithms in the Criminal Justice System, the Law
Society recommended the creation of a register of algorithmic systems in
criminal justice in the UK. The Committee is of the view that such a
register could be expanded beyond criminal justice, if a sensible threshold
is set. In discussions with the Committee, the Centre for Data Ethics and
Innovation expres

   Such a register may be appropriate in other parts of the public
sector.”81 Centre for Data Ethics and Innovation

  However, it is likely that the establishment of a central register, even
if restricted to AI systems above a high threshold, would be an extensive
and potentially overwhelming bureaucratic challenge, particularly given the
predicted scale of AI across public life. T




On Fri, Feb 5, 2021 at 8:43 PM LJ.Garcia <lj.garcia.co@gmail.com> wrote:

> Hi Paola, all,
>
> Thanks for the algorithm watch link. Working on a public schema for a
> registry sounds like a good idea, do you have any ideas on that already?
> What you are proposing is a schema that registries on AI and ADM could
> follow to increase, for instance, transparency, and so other desirable
> characteristics for AI and ADM, am I understanding that correctly?
>
> By the way, what do you call a registry? To me is similar to a directory,
> where you find basic information, possible links to related resources and
> definitely a link to the actual resource (hosted somewhere else). I ask to
> make sure we are all on the same page, I have seen common terms might have
> different interpretations.
>
> These are some of the software registries that I know, not sure those
> follow under the registry category for you.
> * https://bio.tools/
> * https://openebench.bsc.es, a benchmarking approach for for life sciences
> * https://www.european-language-grid.eu/, for NLP and language approaches
> but not publicly open
> * http://gerbil.aksw.org/gerbil/, also for benchmarking
>
> Regards,
>
> On Fri, Jan 29, 2021 at 7:58 AM Paola Di Maio <paoladimaio10@gmail.com>
> wrote:
>
>> LJ
>>
>> KG as KR
>> I hope I can have a discussion with you about the suitability of KG as KR
>> I have developed a checklist for the evaluation of  KR and maybe you can
>> help me
>> future out if KGs meet the adequacy criteria
>>
>> regarding the register, last fall (time goes so quickly) I attended a
>> launch event for
>> a report by algorithm watch
>> https://algorithmwatch.org/en/press-release-automating-society-2020/
>> There they make some recommendations, including for a registry
>>
>> crease the transparency of ADM systems by establishing public registers
>> for ADM systems used within the public sector and by introducing
>> legally-binding data access frameworks to support and enable public
>> interest research;
>>
>> (see the bottom of the link for more recommendations)
>>
>> But I dont think a registry is feasible without massive overhead
>> and centralization,
>>
>>  after bouncing off some ideas here with Carl and Owen who
>> were also thinking registry (not sure in relation to the same initiative)
>> I suggested that working on a public schema for a registry should be
>> feasible for us, ie a set of data/requirements that can be used by public
>> authorities and /or software developers/owners to ensure their AI is
>> trustworthy,  Even just to publish a schema/specification for the registry
>> is going to require commitment
>>
>> regarding stratML , please search our mailing list of extensive
>> references and pointers to resources and if you cannot find
>> them email Owen will point you
>> https://lists.w3.org/Archives/Public/public-aikr/
>>
>> Thank you for joining and I ll soon send a summary of recent published
>> work
>> and talks
>>
>>
>> On Tue, Jan 19, 2021 at 10:11 PM LJ.Garcia <lj.garcia.co@gmail.com>
>> wrote:
>>
>>> Hi Paola, all,
>>>
>>> I work with semantic web and linked data in life sciences so I guess I
>>> am one of those that find knowledge graphs as a good KR. I use semantic
>>> approaches to combine data and do some knowledge discovery / data analytics
>>> there, so I understand better the data and what I can get out of it.
>>>
>>> I have not been able to get much information about StrathML, google
>>> seems to get confused with the term and does not give me much. Any link you
>>> can provide?
>>>
>>> Regarding a schema for ethical AI, I could be interested in it but would
>>> need to understand better what you mean. I have worked with schema.org
>>> to add markup/metadata to webpages. Let's suppose you expose your AI tool
>>> on a portal or so, would the schema you have in mind describe the AI tool
>>> with some properties regarding ethical aspects?
>>>
>>> Regards,
>>>
>>>
>>> On Tue, Jan 19, 2021 at 1:15 AM Paola Di Maio <paoladimaio10@gmail.com>
>>> wrote:
>>>
>>>> LJ,
>>>> please tell us about your topic and your interest in AI KR
>>>>
>>>> I ll be posting some updates soon at my end
>>>> My vision for KR is that is pretty much the center of the known
>>>> universe, in the sense that
>>>> humans relate to the world around them through cognition
>>>> From KR depends everything that is shared/collective, that is made
>>>> explicit and communicated and subject to norms
>>>> That is quite a lot. Now that some level of AI is used in every system
>>>> KR is becoming more important
>>>> But it has not evolved much. I dont think kg (Knowledge Graphs) are
>>>> reliable KR methods. Do you have opinions
>>>> on that?
>>>> I dont need KR in the same way for our own internal world
>>>> this is perhaps more the realm of Cognitive AI
>>>>
>>>> I am pursuing my vision by pitching papers - had a few abstracts and
>>>> papers and talks on the subject
>>>> accepted,  and started this group in the hope the message trickles
>>>> It has trickled, as people have started picking up on the topics we
>>>> share here and discuss them
>>>> elsewhere, for some reason :-)
>>>>
>>>> There is a lot of interesting material and I think I ll start by
>>>> publishing a concept map of what is going on
>>>>
>>>> StrathML is great but fits a bit like a straightjacket for my madness
>>>> at the moment :-)
>>>>
>>>> The last meaningful ideas was a schema for Ethical AI repos written in
>>>> StrathML or something
>>>> which i hope goes ahead
>>>>
>>>>
>>>> pdm
>>>>
>>>> PDM
>>>>
>>>> On Mon, Jan 18, 2021 at 7:07 PM LJ.Garcia <lj.garcia.co@gmail.com>
>>>> wrote:
>>>>
>>>>> Hi Paola,
>>>>>
>>>>> Thanks for your reply. StratML meetings would be too specific for me,
>>>>> great to learn but out of my topic right now to not enough time to engage.
>>>>> I will keep a closer eye to the mailing list to learn more from
>>>>> discussions and so on. If I see I can contribute somehow, I will do so.
>>>>>
>>>>> Thanks!
>>>>>
>>>>> On Thu, Jan 7, 2021 at 3:22 AM Paola Di Maio <paoladimaio10@gmail.com>
>>>>> wrote:
>>>>>
>>>>>> LJ
>>>>>> Thanks for the PIng
>>>>>>
>>>>>> Carl and Owen have been organising meetings focussed on a specific
>>>>>> topic StrathML
>>>>>> but we can set up meetings anytime, with any focus-
>>>>>> Let us know what you work on, what is your interest and vision for
>>>>>> the participation in this group
>>>>>> and we maybe set up a call to focus on any topic that may be of
>>>>>> interest to you and others
>>>>>>
>>>>>> PDM
>>>>>>
>>>>>>
>>>>>> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail> Virus-free.
>>>>>> www.avast.com
>>>>>> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
>>>>>> <#m_-341602378465802529_m_-1310583577550505253_m_1031012630689545055_m_-2155910221626731869_m_4138751991099088002_m_-3762926692016067577_m_-8267902609057381427_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
>>>>>>
>>>>>> On Wed, Jan 6, 2021 at 10:13 PM LJ.Garcia <lj.garcia.co@gmail.com>
>>>>>> wrote:
>>>>>>
>>>>>>> Dear all,
>>>>>>>
>>>>>>> I have partially followed discussions going on on the mailing list
>>>>>>> but this year would like to be more active. I think there is a monthly
>>>>>>> meeting? If yes, could you please remind me the details?
>>>>>>>
>>>>>>> Kind regards,
>>>>>>>
>>>>>>