W3C home > Mailing lists > Public > semantic-web@w3.org > February 2015

Re: [Dbpedia-discussion] [Dbpedia-ontology] Advancing the DBpedia ontology

From: M. Aaron Bossert <mabossert@gmail.com>
Date: Thu, 26 Feb 2015 01:07:37 -0500
Cc: John Flynn <jflynn12@verizon.net>, Linked Data community <public-lod@w3.org>, SW-forum <semantic-web@w3.org>, dbpedia-ontology <dbpedia-ontology@lists.sourceforge.net>, "<dbpedia-discussion@lists.sourceforge.net>" <dbpedia-discussion@lists.sourceforge.net>
Message-Id: <47581E14-3C13-482B-9274-3ACDD3EE457B@gmail.com>
To: Mike Bergman <mike@mkbergman.com>
The one thing I would say is that while I agree in general...the one thing that keeps eating away at me is that there is tremendous potential in dbpedia for bigger questions to be answered, but the more advanced analytics require that some level of sanity exists within the ontology...much more so than now.  As an example, I have created several different applications for customers that are based on dbpedia...one of which is a recommender system.  The level of effort required to simply say (in SPARQL, of course) "show me every living person that is highly similar to person X, excluding politicians athletes and actors" is quite a tedious thing to do until after I have "fixed" all the erroneous and missing properties associated with "things" in general...which person class do I focus on?  Which living people?  Which politicians?  Perhaps legislators?  It gets pretty ugly, pretty quickly.

I'm not sure that the ontology needs to be completely rewritten, but surely it can't be that difficult to clean up a bit with a little common sense logic applied such as if a "thing" has a death date (never mind which one), then surely they are not a living person...or if they hold a political office, surely they must be a politician.

Aaron

> On Feb 26, 2015, at 00:19, Mike Bergman <mike@mkbergman.com> wrote:
> 
> Hi John,
> 
> My thoughts are for DBpedia to stay close to the mission of extracting quality data from Wikipedia, and no more. That quality extraction is an essential grease to the linked data ecosystem, and of much major benefit to anyone needful of broadly useful structured data.
> 
> I think both Wikipedia and DBpedia have shown that crowdsourced entity information and data works beautifully, but the ontologies or knowledge graphs (category structures) that emerge from these effort are mush.
> 
> DBpedia, or schema.org from that standpoint, should not be concerned so much about coherent schema, computable knowledge graphs, ontological defensibility, or any such T-Box considerations. They have demonstrably shown themselves to not be strong in these suits.
> 
> No one hears the term "folksonomy" any more because all initial admirers have seen no crowd-sourced schema to really work (from dmoz to Freebase). A schema is not something to be universally consented, but a framework by which to understand a given domain. Yet the conundrum is, to organize anything globally, some form of conceptual agreement about a top-level schema is required.
> 
> Look to what DBpedia now does strongly: extract vetted structured data from Wikipedia for broader consumption on the Web of data.
> 
> My counsel is to not let DBpedia's mission stray into questions of conceptual "truth". Keep the ontology flat and simple with no aspirations other than "just the facts, ma'am".
> 
> Thanks, Mike
> 
>> On 2/25/2015 10:33 PM, M. Aaron Bossert wrote:
>> John,
>> 
>> You make a good point...but are we talking about a complete tear-down of the existing ontology?  I'm not necessarily opposed to that notion, by want to make sure that we are all in agreement as to the scope of work, as it were.
>> 
>> What would be the implications of a complete redo?  Would the benefit outweigh the impact to the community?  I would assume that there would be a ripple effect across all other LOD datasets that map to dbpedia, correct?  Or am I grossly overstating/misunderstanding how interconnected the ontology is?
>> 
>> Vladimir, your thoughts?
>> 
>> Aaron
>> 
>>> On Feb 25, 2015, at 21:14, John Flynn <jflynn12@verizon.net> wrote:
>>> 
>>> It seems the first level effort should be a requirements analysis for the
>>> Dbpedia ontology.
>>> - What is the level of expressiveness needed in the ontology language- 1st
>>> order logic, some level of descriptive logic, or a less expressive language?
>>> - Based on the above, what specific ontology implementation language should
>>> be used?
>>> - Should the Dbpedia ontology leverage an existing upper ontology, such as
>>> SUMO, DOLCE, etc?
>>> - Should the Dbpedia ontology architecture consist of a basic common core of
>>> concepts (possibly in addition to the concepts in a upper ontology) that are
>>> then extended by additional domain ontologies?
>>> - How will the Dbpedia ontology be managed?
>>> - What are the hosting requirements for access loads on the ontology? How
>>> many simultaneous users?
>>> 
>>> This is only a cursory cut at Dbpedia ontology requirement issues. But, it
>>> seems the community needs to come to grips with this issue before
>>> implementing specific changes to the existing ontology.
>>> 
>>> John Flynn
>>> http://semanticsimulations.com
>>> 
>>> -----Original Message-----
>>> From: M. Aaron Bossert [mailto:mabossert@gmail.com]
>>> Sent: Wednesday, February 25, 2015 9:13 AM
>>> To: <vladimir.alexiev@ontotext.com>
>>> Cc: dbpedia-ontology; Linked Data community; SW-forum;
>>> <dbpedia-discussion@lists.sourceforge.net>
>>> Subject: Re: [Dbpedia-ontology] [Dbpedia-discussion] Advancing the DBpedia
>>> ontology
>>> 
>>> Vladimir,
>>> 
>>> I'm thinking of trying to do some stats on the existing ontology and the
>>> mappings to see where there is room for improvement.  I'm tied up this week
>>> with a couple deadlines that I seem to moving towards at greater than light
>>> speed, though my progress is not.
>>> 
>>> As soon as I get the rough cut done, I'll share the results with you and
>>> maybe we can discuss paths forward?
>>> 
>>> I'm with you on the 30% error rate...that doesn't help anyone.
>>> 
>>> Aaron
>> 
>> ------------------------------------------------------------------------------
>> Dive into the World of Parallel Programming The Go Parallel Website, sponsored
>> by Intel and developed in partnership with Slashdot Media, is your hub for all
>> things parallel software development, from weekly thought leadership blogs to
>> news, videos, case studies, tutorials and more. Take a look and join the
>> conversation now. http://goparallel.sourceforge.net/
>> _______________________________________________
>> Dbpedia-discussion mailing list
>> Dbpedia-discussion@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
> 
> 
Received on Thursday, 26 February 2015 06:08:14 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 08:45:41 UTC