GAO Antifraud Resource released

21 views
Skip to first unread message

Dickerson, Leia J

unread,
Jan 13, 2022, 9:59:51 AM1/13/22
to ontolo...@googlegroups.com, antifraud

Hello all—

I’m writing to provide an update on a project I presented on during last year’s Ontology Summit.

 

Released on Monday, the GAO Antifraud Resource (GAO Antifraud Resource) is based on a GAO’s Fraud Ontology—a rigorous classification of fraud schemes affecting federal programs and operations—serves as the backbone for understanding, evaluating, and measuring all aspects of federal fraud schemes, including their participants, mechanisms, and impacts.

 

The site provides a user friendly, web-based platform for interacting with the model and identifying resources to support fraud risk management.

 

The ontology is in its very first version.  We plan to edit it and add to it (including modularizing it) in future iterations.  We welcome your feedback; please send it to anti...@gao.gov.

 

Thank you.

 

Leia Dickerson | she/her/hers | pronounced lee-uh

Senior Librarian, Taxonomy + Research | Center for Library Science

Applied Research + Methods | US Government Accountability Office

Dicke...@gao.gov | https://www.gao.gov

 

Azamat Abdoullaev

unread,
Jan 13, 2022, 10:27:42 AM1/13/22
to ontolo...@googlegroups.com
Leia,
Wonder if AI, ML, DL, or Data Analytics tools are used in fraud classification and detection
Thanks
image.png

--
All contributions to this forum are covered by an open-source license.
For information about the wiki, the license, and how to subscribe or
unsubscribe to the forum, see http://ontologforum.org/info/
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ontolog-forum/PH0PR09MB839483BF78BB45247B1E7F7CA5539%40PH0PR09MB8394.namprd09.prod.outlook.com.

Dickerson, Leia J

unread,
Jan 13, 2022, 5:48:15 PM1/13/22
to ontolo...@googlegroups.com, antifraud

Thank you for your question, Azamat.

 

We did not use AI, machine learning, or deep learning in creating this resource.  We did use content analysis, but not other data analysis.

 

Still, we are interested in employing such tools for our future work.

 

From: ontolo...@googlegroups.com <ontolo...@googlegroups.com> On Behalf Of Azamat Abdoullaev
Sent: Thursday, January 13, 2022 10:27 AM
To: ontolo...@googlegroups.com
Subject: Re: [ontolog-forum] GAO Antifraud Resource released

 

CAUTION EXTERNAL EMAIL: Do not click on any links or open any attachments unless you trust the sender and/or know the content is safe. If you are suspicious of the e-mail, click on the Report Suspicious Emails button.

Azamat Abdoullaev

unread,
Jan 14, 2022, 2:27:15 AM1/14/22
to ontolo...@googlegroups.com
Leia,
I saw your Conceptual Fraud Model (CoFM), which looks rather engaging. 
A couple of free suggestions. 
The CoFM had better transform into a Causal FM (CaFM) with all possible relationships among agents (not Knowledge Graphs, but Causal Network Graphs).
To animate the CaFM, it needs to be intelligently automated by applying AI/ML/DL/DA techniques. 
Data analytics - for prediction, Machine Learning to process structured data, Deep Learning - unstructured data, and AI models for reasoning and learning.
As I understand, there are multi-billion fraud datasets which could be used for training and testing and validating the AI Fraud Models
In the good business and government practice, the ML fraud detection algorithms save billions for insurance and banking sectors, even in the conceptually deficient model-free ML version.
Today, Causal/Real AI is not a "nice-to-have" thing, but "must-have", if it is to survive and prosper.  
Certainly, considering the scale and scope of your project, it must be the Causal AI/ML/DL, and not today's Statistic AI/ML/DNNs.
Last, not least, the whole enterprise could be advanced as the i-America initiative. 
Regards,
.

To view this discussion on the web visit https://groups.google.com/d/msgid/ontolog-forum/PH0PR09MB839439377C614D35DA3924C7A5539%40PH0PR09MB8394.namprd09.prod.outlook.com.

John Bottoms

unread,
Jan 14, 2022, 7:31:59 PM1/14/22
to ontolo...@googlegroups.com

AA,

Sorry, I missed a definition, "Casual AI".

Is that a tongue in cheek oxymoron? Or could you explain what can be casual about intelligence?

-John Bottoms, firstStarSystems

John Bottoms

unread,
Jan 15, 2022, 1:39:04 AM1/15/22
to ontolo...@googlegroups.com

AA,

Sorry, I read that without cleaning my glasses. Causal reasoning could be considered a restatement of other types of analysis that predate any connection to AI. The most well known is Hedonistic Analysis which is a human curated analysis of the types of events that could affect a model, discarding those that were obviously non-causal. These issues are sometimes linked to Spillover which examines causal events across time. The writings by "Komarkova, Conrads, & Collado, 2015" are the first I have seen that lay out the methodology and math behind spillover. Earlier analysis were done by hand.

It seems to me that any attempt to automate causal reasoning without human input is potentially biased. These appear to be system level issues that are not strictly a single domain issue.

-John Bottoms, FirstStar Systems


On 1/14/2022 2:26 AM, Azamat Abdoullaev wrote:

Azamat Abdoullaev

unread,
Jan 15, 2022, 9:00:12 AM1/15/22
to ontolo...@googlegroups.com
JB:  
Causal reasoning could be considered a restatement of other types of analysis that predate any connection to AI. The most well known is Hedonistic Analysis which is a human curated analysis of the types of events that could affect a model, discarding those that were obviously non-causal. These issues are sometimes linked to Spillover which examines causal events across time. The writings by "Komarkova, Conrads, & Collado, 2015" are the first I have seen that lay out the methodology and math behind spillover. Earlier analysis were done by hand.

It is all about causal nexus, as the full causal network of n-causes-effects and n (n-1)/2 causal connections. It includes spillover effects and network effects as its special topologies. 
For example, the larger the power of an economy is, the more spillover effects it is likely to produce across the global economy.
Most of the world experiences significant spillover effects when there is a downturn or macro effect in the world's two largest economies: the United States and China.
For telecommunications networks, the internet, web, social media, the phenomenon described as Metcalfe's Law that the value or utility of a service, product or technology is proportional to the number of connected users of the system (n2).

JB: It seems to me that any attempt to automate causal reasoning without human input is potentially biased.
In fact, causal AI automates causal reasoning (causality identification, types of causal relations, or inferring cause and effect in the datasets) without any human intervention; for humans bring in a lot of cognitive biases, as it happens with today's ML/algorithm/AI bias, racial, gender, etc. coming from incomplete, faulty or prejudicial data sets to train and/or validate the machine learning systems. Instead of GIGO, we go for CICO (cause in, cause out).
In the RSI machines, causal understanding is automatic, like in humans, it is intuitive or insightful.

To view this discussion on the web visit https://groups.google.com/d/msgid/ontolog-forum/7744dd29-2dda-0db0-6dda-8a03b3c4d6f6%40firststarsystems.com.

Sheth, Amit

unread,
Jan 29, 2022, 2:37:07 PM1/29/22
to ontolo...@googlegroups.com

Nancy Wiegand

unread,
Jan 29, 2022, 3:04:06 PM1/29/22
to ontolo...@googlegroups.com, Nancy Wiegand
Not to comment on OWL or the article as a whole, but to remark on one section:

Regarding the statement in the article: "Logically, saying that Alice is Bob’s sister implies that Bob is Alice’s brother. These are not two independent facts, but logically equivalent ways of stating the same fact. Deriving one from the other does not yield new information, but only an alternative way of conveying the same information.",

one could say that related information, of course, always is there somewhere. It's just that such relationships might not be obvious to a person (maybe never and not always straightforwardly) and certainly not obvious to a computer that takes all terms literally.
--
All contributions to this forum are covered by an open-source license.
For information about the wiki, the license, and how to subscribe or
unsubscribe to the forum, see http://ontologforum.org/info/
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.

Pascal Hitzler

unread,
Jan 29, 2022, 3:22:15 PM1/29/22
to ontolo...@googlegroups.com
Actually,

"Logically, saying that Alice is Bob’s sister implies that Bob is
Alice’s brother. These are not two independent facts, but logically
equivalent ways of stating the same fact. "

is just plain wrong. Alice is Bob's sister implies that Bob is Alice's
sibling, but not brother.

At this stage I'm wondering whether I should in fact even read on, given
that this signals a certain limited understanding of formal logic (which
is the basis of OWL) :)

Pascal.

On 1/29/2022 2:04 PM, Nancy Wiegand wrote:
> Not to comment on OWL or the article as a whole, but to remark on one
> section:
>
> Regarding the statement in the article: "Logically, saying that Alice is
> Bob’s sister implies that Bob is Alice’s brother. These are not two
> independent facts, but logically equivalent ways of stating the same
> fact. Deriving one from the other does not yield new information, but
> only an alternative way of conveying the same information.",
>
> one could say that related information, of course, always is there
> somewhere. It's just that such relationships might not be obvious to a
> person (maybe never and not always straightforwardly) and certainly not
> obvious to a computer that takes all terms literally.
>
>
> On 1/29/2022 1:37 PM, Sheth, Amit wrote:
>> This may generate some discussion here:
>>
>> https://www.linkedin.com/pulse/syntax-semantics-great-owl-hoax-jan-voskuil
>>
>>
>>
>>
>> AIISC <http://ai.sc.edu/>
>> *Amit Sheth <mailto:am...@sc.edu>* • LinkedIn
>> <http://www.linkedin.com/in/amitsheth> • Google Scholar
>> <https://scholar.google.com/citations?user=2T3H4ekAAAAJ&hl=en> • Quora
>> <https://www.quora.com/profile/Amit-Sheth-1> • Blog
>> <http://amitsheth.blogspot.com/> • Twitter <http://twitter.com/amit_p>
>> Founding Director, Artificial Intelligence Institute (#AIISC)
>> <http://j.mp/UofSCAINews>
>> University of South Carolina (UofSC)
>> #AIISC on LinkedIn <https://www.linkedin.com/company/1054055/>• #AIISC
>> on FB <https://www.facebook.com/AIIatUofSC/>
>> #AIISC on Web <http://ai.sc.edu/>• #AIISC News <http://j.mp/UofSCAINews>
>>
>>
>> --
>> All contributions to this forum are covered by an open-source license.
>> For information about the wiki, the license, and how to subscribe or
>> unsubscribe to the forum, see http://ontologforum.org/info/
>> ---
>> You received this message because you are subscribed to the Google
>> Groups "ontolog-forum" group.
>> To unsubscribe from this group and stop receiving emails from it, send
>> an email to ontolog-foru...@googlegroups.com.
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/ontolog-forum/C0B5F3E5-3E2B-44F6-9889-29EAAD0B7C67%40sc.edu
>> <https://groups.google.com/d/msgid/ontolog-forum/C0B5F3E5-3E2B-44F6-9889-29EAAD0B7C67%40sc.edu?utm_medium=email&utm_source=footer>.
>
>
> --
> All contributions to this forum are covered by an open-source license.
> For information about the wiki, the license, and how to subscribe or
> unsubscribe to the forum, see http://ontologforum.org/info/
> <http://ontologforum.org/info/>
> ---
> You received this message because you are subscribed to the Google
> Groups "ontolog-forum" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to ontolog-foru...@googlegroups.com
> <mailto:ontolog-foru...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/ontolog-forum/ddbc2e04-5ee3-72c3-39e5-44b6ee14a9ae%40cs.wisc.edu
> <https://groups.google.com/d/msgid/ontolog-forum/ddbc2e04-5ee3-72c3-39e5-44b6ee14a9ae%40cs.wisc.edu?utm_medium=email&utm_source=footer>.

--
Pascal Hitzler
Lloyd T. Smith Creativity in Engineering Chair
Director, Center for AI and Data Science
Kansas State University http://www.pascal-hitzler.de
http://www.daselab.org http://www.semantic-web-journal.net

doug foxvog

unread,
Jan 29, 2022, 6:38:07 PM1/29/22
to ontolo...@googlegroups.com
This article discusses some of the problems with OWL, but accepts RDF
triple syntax, which seems to me a mistake. A useful language would allow
ternary, quaternary, and variable-arity predicates.

I also find a problem with
" The notion of class implies a notion of individual. OWL makes a point
of formalizing this distinction. As long as you make sure your classes
and individuals are disjunct [disjoint?], OWL reasoning is efficient.
...
However, OWL does not enforce it. It is probably best to treat this as a
principle of ontological hygiene."

Yes, there is a distinction between Class and Individual -- they are
disjoint classes. An instance of Class can itself have instances, while
an instance of Individual can not.

The notion of a class implies the notion of an INSTANCE, not of
INDIVIDUAL. Instancehood is a property, not really a class, because one
can make any class an instance of another class. Note that one can have
classes at different levels:
The class BiologicalTaxonType has the class BiologicalSpecies as an instance.
The class BiologicalSpecies has the class HomoSapiens as an instance.
The class HomoSapiens has the individual DougFoxvog as an instance.
The individual DougFoxvog can have no instances since it is not a class.

Note that one can also have classes of predicate, such as BinaryPredicate,
ReflexiveBinaryPredicate, TransitiveBinaryPredicate, an
QuaternaryPredicate.

-- doug foxvog


On Sat, January 29, 2022 14:37, Sheth, Amit wrote:
> This may generate some discussion here:
>
> https://www.linkedin.com/pulse/syntax-semantics-great-owl-hoax-jan-voskuil
>
>
>
> [AIISC] <http://ai.sc.edu/>
> Amit Sheth<mailto:am...@sc.edu> •
> LinkedIn<http://www.linkedin.com/in/amitsheth> • Google
> Scholar<https://scholar.google.com/citations?user=2T3H4ekAAAAJ&hl=en> •
> Quora<https://www.quora.com/profile/Amit-Sheth-1> •
> Blog<http://amitsheth.blogspot.com/> •
> Twitter<http://twitter.com/amit_p>
> Founding Director, Artificial Intelligence Institute
> (#AIISC)<http://j.mp/UofSCAINews>
> University of South Carolina (UofSC)
> #AIISC on LinkedIn<https://www.linkedin.com/company/1054055/>• #AIISC on
> FB<https://www.facebook.com/AIIatUofSC/>
> #AIISC on Web<http://ai.sc.edu/>• #AIISC News<http://j.mp/UofSCAINews>
>

Alex Shkotin

unread,
Jan 30, 2022, 2:38:11 AM1/30/22
to ontolog-forum
Pascal,

After reading just an abstract let me say that it is about SHACL vs OWL with maxima "we can live without OWL":-)

Alex

сб, 29 янв. 2022 г. в 23:22, 'Pascal Hitzler' via ontolog-forum <ontolo...@googlegroups.com>:
To view this discussion on the web visit https://groups.google.com/d/msgid/ontolog-forum/ae88f604-1a6c-0f20-615d-d25fce8ab5d5%40googlemail.com.

Alex Shkotin

unread,
Jan 30, 2022, 2:46:28 AM1/30/22
to ontolog-forum
just more citation: " SHACL reasoners also generate triples, and therefore generate “new facts” in a way similar to OWL reasoners. SHACL reasoners, however, do not apply rules of logic. The rules they execute apply custom defined shapes. These will then generate triples with surgical precision, in a controlled fashion."

What do we know about SHACL as onto-language. Do we have it in the DOL?

Alex


сб, 29 янв. 2022 г. в 22:37, Sheth, Amit <AM...@sc.edu>:
--
All contributions to this forum are covered by an open-source license.
For information about the wiki, the license, and how to subscribe or
unsubscribe to the forum, see http://ontologforum.org/info/
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ontolog-forum/C0B5F3E5-3E2B-44F6-9889-29EAAD0B7C67%40sc.edu.

Alex Shkotin

unread,
Jan 30, 2022, 3:16:35 AM1/30/22
to ontolog-forum
Amit,

Thank you. The question about OWL2 usage including reasoners is very interesting. And discussion is very interesting too.
Time to study SHACL:-)

Alex

сб, 29 янв. 2022 г. в 22:37, Sheth, Amit <AM...@sc.edu>:
This may generate some discussion here:

--
All contributions to this forum are covered by an open-source license.
For information about the wiki, the license, and how to subscribe or
unsubscribe to the forum, see http://ontologforum.org/info/
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ontolog-forum/C0B5F3E5-3E2B-44F6-9889-29EAAD0B7C67%40sc.edu.

alex.shkotin

unread,
Jan 30, 2022, 3:32:55 AM1/30/22
to ontolog-forum
Is that a good idea to ask Jan about SWRL?

суббота, 29 января 2022 г. в 22:37:07 UTC+3, Sheth, Amit:

Igor Toujilov

unread,
Jan 30, 2022, 3:02:49 PM1/30/22
to ontolo...@googlegroups.com
I strongly disagree. There is no deep analysis in this article. Most statements there are unjustified. The main argument based on the date of birth integrity constraint easily collapses when you know how to do it properly in OWL: have a look at https://sourceforge.net/p/meloproject/wiki/open_world_and_closed_world/ , section "The Common Myth: OWL Does not Support Integrity Constraints because of the Open World". And you can do almost everything you need in OWL2+SWRL.
So, instead of denying OWL, it is much more productive to learn how to use OWL and related technologies properly.

--
All contributions to this forum are covered by an open-source license.
For information about the wiki, the license, and how to subscribe or
unsubscribe to the forum, see http://ontologforum.org/info/
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ontolog-forum/C0B5F3E5-3E2B-44F6-9889-29EAAD0B7C67%40sc.edu.

Robert Goldman

unread,
Jan 31, 2022, 9:48:04 AM1/31/22
to ontolog-forum
Could we move this to a different thread? It's not good that a discussion of this article is buried in a discussion about a GAO Fraud ontology.

I don't know how well this can work, since I don't know much about Google Groups. Ideal would be to just move this discussion out of this thread, if an admin could do that. Less ideal, but still better than continuing as we are would be to just discuss this in a fresh thread, possibly started by a first message that is a digest of the discussion so far.

Alex Shkotin

unread,
Jan 31, 2022, 11:20:54 AM1/31/22
to ontolog-forum
Robert,

I agree and one simple way is to begin a new thread and just put reference to Amit's post [1] from this thread.
Everybody can do this but who can digest:-)

Alex

[1] https://groups.google.com/g/ontolog-forum/c/hkXFMzkr-ks/m/5xpbMXqhAQAJ

пн, 31 янв. 2022 г. в 17:48, Robert Goldman <rpgo...@gmail.com>:
To view this discussion on the web visit https://groups.google.com/d/msgid/ontolog-forum/1652e042-f52e-490d-8558-316e50d3a854n%40googlegroups.com.
Reply all
Reply to author
Forward
0 new messages