ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Toward Human-Level Artificial Intelligence

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: Kingsley Idehen <kidehen@xxxxxxxxxxxxxx>
Date: Mon, 28 Apr 2014 08:37:17 -0400
Message-id: <535E4B7D.9020100@xxxxxxxxxxxxxx>
On 4/27/14 5:12 PM, John F Sowa wrote:
> Kingsley,
>
> I agree with you that DBpedia is a very large database with a very
> large amount of information that can support a significant amount
> of question answering and reasoning about the answers.
>
> I also agree that Schema.org is a hierarchy of terms that do not
> contain anything that resembles the facts contained in DBpedia.
>
> But note that Schema.org contains a lot OWL-level specifications,
> including the GoodRelations ontologies.  That is the point I was
> making in the following statement:
>
> JFS
>>> The DBpedia terms are at the same level of vagueness as Schema.org.    (01)

They aren't . GoodRelations assimilation does have the downside of its 
semantics being blurred. Fundamentally, Schema.org is a vocabulary with 
absolute minimal semantic fidelity. In short, GoodRelations assimilation 
is done by way of complete namespace replacement without any use of the 
following:    (02)

1. owl:equivalentClass
2. owl:equivalentProperty
3. rdfs:subClassOf
4. rdfs:subPropertyOf
5. owl:inverseOf.    (03)

Basically, you have to post-process, as I do, to make sense of data 
constructed using Schema.org terms, in regard to actual relation semantics.    (04)

Schema.org is vague for sure, and that lies at the core of its design. 
On the other hand, the DBpedia ontology isn't vague, for starters it 
handles unambiguous entity denotation in regards to:    (05)

1. class definition
2. relation definition
3. relationship construction .
> KI
>> No they aren't. You can reason using the DBpedia ontology, you can't
>> using Schema.org without using some post processing to fix all its
>> denotation related ambiguities.
> Could you give some examples of similar terms -- as specified
> in DBpedia and in Schema.org.    (06)

Great question! On the DBpedia side, there are individual efforts by 
those that are interested in building bridges [1][2].    (07)

Irrespective of how the mapping is done, you need to apply an inference 
and reasoning context based on at least one of the following relations:    (08)

• owl:equivalentClass
• owl:equivalentProperty
• rdfs:subClassOf
• rdfs:subPropertyOf
• owl:inverseOf.    (09)

And before you embark on the above, you need to fix the Schema.org 
entity denotation, which (as I've stated repeatedly) conflates the 
identities of documents and the entities they describe.    (010)

>   And show what post processing you
> need to do with the Schema.org terms to resolve the ambiguities.    (011)

I post-process Schema.org by indicating to my tool (Virtuoso) that 
certain Schema.org classes are equivalent to classes on other 
ontologies, and then use the relation semantics in the other ontologies 
to perform reasoning, since its those other ontologies that provide the 
actual entity relation semantics that are comprehensible to my tool 
(which isn't a human being reading text written in English).    (012)

Schema.org makes the assumption that you are an English speaker that 
will read paragraphs, then go away and write code in line with what it 
describes. As you know, the ontologies we are talking about circa., 
2014, are those that are comprehensible to humans and machines i.e., 
they don't depend on a human being that speaks, reads, and thinks in 
English.    (013)

As you've espoused, It should be all about logic, which is a universal 
language.    (014)

Links:    (015)

[1] 
http://lod.openlinksw.com/describe/?url=http%3A%2F%2Fschema.org%2FLandmarksOrHistoricalBuildings%23this&graph=urn%3Acom%3Aopenlinksw%3Amappings    (016)

-- example that demonstrates how "#this" is used to make unambiguous 
entity names from schema.org names as part of an equivalentClass mapping 
to a DBpedia Class (without this, one is basically asserting 
inaccurately that a document denoted by an HTTP URL and a class denoted 
by an HTTP URI are one and the same, which it utterly wrong when 
reasoning and inference is in play)    (017)

[2] 
http://lod.openlinksw.com/describe/?url=http%3A%2F%2Fschema.org%2FalumniOf%23this&graph=urn%3Acom%3Aopenlinksw%3Amappings    (018)

-- another example    (019)

[3] https://plus.google.com/+KingsleyIdehen/posts/QbGxuMuiJ9J -- 
detailed post related to the items above    (020)

[4] http://bit.ly/1m07ZXn -- SPARQL query results that unveils the 
relations used in this particular DBpedia to Schema.org mappings    (021)

[5] http://bit.ly/1itAwR0 -- SPARQL query definition behind the results 
page above.    (022)

>
> John
>   
>    (023)


--     (024)

Regards,    (025)

Kingsley Idehen 
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter Profile: https://twitter.com/kidehen
Google+ Profile: https://plus.google.com/+KingsleyIdehen/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen    (026)

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature


_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)

<Prev in Thread] Current Thread [Next in Thread>