Re: Fwd:

Hi Paola,

The KM system Peter Clark worked on is quite nice. The AURA system used 
only a fraction of features of KM. The features of KM that are used in 
AURA were formalized to some degree of coverage in this paper: 
http://www.ai.sri.com/pubs/files/1958.pdf.   There are several other 
papers related to KRR of AURA that we wrote when Project Halo ended.

We are still trying to see if we can find enough adoption of an end-user 
app that we had built based on AURA. That effort is still active. See 
http://web.stanford.edu/~vinayc/intelligent-life/ for its current status.

A serious problem with KM was its destructive implementation: when it 
derived new inferences, it overwrote user-entered information. This 
engineering flaw made KM un-viable.

As you will see from the first paper cited above, AURA uses a language 
that is equivalent to existential rules. Substantial theory exists 
formalizing such rules. E.g, see the following works 
http://www.lirmm.fr/~mugnier/graphik/kiabora/downloads/framework_en.pdf 
and https://www.sciencedirect.com/science/article/pii/S0004370211000397

With some luck, we will rebuild AURA using this new theory and more 
recent results on existential rules.  We are also contemplating a 
successor of KM that has potential new trajectories that were not 
considered in AURA.

Regarding your question about what other KR languages are being used, 
here is some additional information:

At Stanford, Michael Genesereth has developed a system called Epilog 
that he uses in several of his courses. Epilog is safe-stratified 
Datalog.  You can find his language described at: 
http://worksheets.stanford.edu/documentation/reference.php

Bob Kowalski has a system called LPS that uses a version of production 
of system for modeling contracts. For example, see: http://lps.doc.ic.ac.uk/

In addition, you may want to reach out to Sheila McIlraith  and Michael 
Kifer.  They both have active interest in this topic.

I am happy to collaborate with you on this further.

Thanks,

Vinay



On 10/10/2018 7:14 PM, Paola Di Maio wrote:
> I found a working note from Peter Clarke
> Ihttp://www.cs.utexas.edu/users/pclark/working_notes/010.pdf
> so I got in touch to ask him if he would give us some contribution he 
> sent some interesting input
>
>
> Hi Paola –
>
> Thanks for your interest! Yes, that’s an old paper – 1996 (the full 
> list of those working notes is here 
> <http://www.cs.utexas.edu/users/pclark/working_notes/>) – I’m glad you 
> found it interesting! It’s been so long since I wrote it, it was quite 
> interesting to me to read through again. It was written at the time to 
> help map the way forward for the KR language Algernon.
>
> Unfortunately I have too many other commitments to establish a 
> collaboration, but I hope your survey goes well! The KR language I 
> have been most involved in and was a primary author of, which (rather 
> biasedly I like a lot!) is KM 
> <http://www.cs.utexas.edu/users/mfkb/km.html>, although it’s no longer 
> in use – it was the primary KR language used in the AURA 
> <https://www.sri.com/work/projects/aura> system (also now defunct), 
> part of Vulcan’s project Halo 
> <https://pdfs.semanticscholar.org/c93c/d0fb5e03a6c931d4e5f48013c3448ff7c733.pdf> 
> (now ended).
>
> Also you might be interested in a A-Z of KR topics 
> <http://www.cs.utexas.edu/users/mfkb/related.html> I maintained until 
> about 5-10 years ago here.
>
> Best wishes,
>
>    Pete
>
> ---
>
> Peter Clark (peterc@allenai.org <mailto:peterc@allenai.org>) tel: 
> (206) 548-5628
>
> Allen Institute for AI,    2157 North Northlake Way, Seattle, WA 98103
>
> http://www.allenai.org (Personal URL: http://allenai.org/team/peterc)
>
> *From:* Paola Di Maio <paola.dimaio@gmail.com 
> <mailto:paola.dimaio@gmail.com>>
> *Sent:* Saturday, October 6, 2018 11:20 PM
> *To:* peterc@allenai.org <mailto:peterc@allenai.org>
> *Subject:*
>
> Dear Peter
>
> I am the chair of a community group AI KR
>
> https://www.w3.org/community/aikr/
>
> trying to survey the current KR languages in use in AI learning/teaching
>
> in particular  natural language representation vs other notations
>
> I found your interesting working note
>
> http://www.cs.utexas.edu/users/pclark/working_notes/010.pdf
>
>
> how old is the paper?(could not find a date)
>
> any updates?
>
> is there a way of establishing collaboration with you on this topic 
> via our group?
>
> can I invite you to participate in an online discussion with our group
>
> if yhou have one hour
>
> Thanks
>
> Best regards
>
>
> Dr Paola Di Maio
>
> NCKU
>
> *A bit about me <https://about.me/paoladimaio>*
>