- From: Richard Wallis <rjw@dataliberate.com>
- Date: Wed, 20 Jun 2018 11:29:28 +0100
- To: Jeff Thompson <jefft0@gmail.com>
- Cc: public-schemaorg@w3.org
- Message-Id: <A043E900-F269-4DB1-A941-69448A6F12BA@dataliberate.com>
Little did I know that my innocent “I don’t see it that way” comment before retiring for the night, would result in me being greeted in the morning with an avalanche of Semantic Web and ontological theory and philosophy that serves to confuse me even more.
Don’t get me wrong, I respect and am sometimes in awe of the forensic focus some can apply to the detail of these discussions/arguments.
However, I do wonder how much relevance this all has to those looking to apply the [schema.org] vocabulary to increase the visibility, discoverability, and general usefulness of their products, services, information, and cultural heritages resources.
I have spent over a decade helping people and organisations pragmatically apply linked data principles and vocabularies such as Schema.org. In that time I would characterise those I have helped as overwhelmingly, either unaware or, perhaps more significantly, not caring where a type came from and what semantic relationship it has with other types. They just want to find the appropriate type for the thing they are describing, follow some examples, and move on to other productive efforts.
Someone earlier in this thread suggested that the problem may be with the English language. Unfortunately, through accidents of history and technological spread, English is the language used by the thousands of developers applying the vocabulary for benefit. That is the English of the street and of business, not of semantic and ontological debate.
Is Schema.org perfect? Far from it.
Is it inconsistent? Yes.
Is it uneven in is coverage of some domains? Yes.
Could it do with better documentation? Most definitely.
Should it have more and varied examples? Hell yes!
Yet despite all of its faults, I believe it can be credited with contributing more to practically realising Sir Tim Berners Lee’s vision of a Semantic Web, with all entities interconnected and described in data, than most anything that has gone before.
Like Thad I’m probably going to move on from this debate to focus on practical and pragmatic adjustments to the vocabulary of terms and properties to improve the visibility of things on 10s of millions of sites and 100s of millions of pages.
As a final comment I would respectfully ask us all to remember who are the target audience for Schema.org (the website developer community) and what their day-to-day use cases are. Many new to Schema.org, looking to this list for help and insight, are probably loosing the will to go on right now.
~Richard.
Sent from my iPhone
> On 20 Jun 2018, at 08:44, Jeff Thompson <jefft0@gmail.com> wrote:
>
> The lack of progress in this conversation shows the limitation of the static propositional view of knowledge representation taken by the semantic web. The teams that are making real progress have abandoned the static approach to modeling knowledge and realize that a model is a program that a cognitive system executes to explain some data. That is the semantics. Without specifying how the program will be executed with some sample data, there is no knowledge representation. For example, here is what Gamalon says who has the world's state-of-the-art language representation system: "models should be programs"
> https://gamalon.com/technology/
>
> Trying use static ontologies and triples to do something more sophisticated that tabulating data will fail (as this thread and the lack of progress over the decades has shown).
>
> - Jeff
>
>> On 2018/06/20 5:48, Sebastian Samaruga wrote:
>> Events denoting dimensional changes of State (measures / observations) over an entity? Periods / processes: spans while an observation in respect to a (dimensional) concept remains the same?
>>
>> State: countryPresident,
>> Event: presidentialElection,
>> Period / Process: presidentialRuling.
>>
>> Still thinking in a more semiotical / dimensional approach.
>>
>>> On Tue, Jun 19, 2018, 10:58 PM Sebastian Samaruga <ssamarug@gmail.com> wrote:
>>> Semiotically, an Event could be regarded as the occurrence of a Concept in the form of a Sign for a given Object (metaclass / class / instance relationship). Occurrences of entities happens as dimensional measures, not only temporal or physical ('birthdays' could be a 'dimension'). Measures bring Data from which Contexts (schema) could be aggregated. And from schema one could infer Interactions or roles in behaviors (DCI OO design pattern).
>>>
>>> This is roughly sketched at:
>>> https://github.com/ssamarug/ssamarug/blob/master/Metamodel.pdf?raw=true
>>>
>>> I think dimensional modelling approach is not only useful in OLAP cubes but also in the Semantic Web. For example in this Event / Period dilemma.
>>>
>>> Is this discussion pertinent to the SW community in general (semantic-web)? As this is an ontological issue I'm including the corresponding list. Regards,
>>>
>>> Sebastián.
>>> http://exampledotorg.blogspot.com
>>>
>>>
>>>> On Tue, Jun 19, 2018, 10:38 PM <Simon.Cox@csiro.au> wrote:
>>>> That is signified by the fact that '19 June 2018' doesn't have a name, only an index.
>>>>
>>>> An 'event' that happened on that day would probably have a name relating to the activity.
>>>>
>>>> -----Original Message-----
>>>> From: Peter F. Patel-Schneider [mailto:pfpschneider@gmail.com]
>>>> Sent: Wednesday, 20 June, 2018 11:17
>>>> To: public-schemaorg@w3.org
>>>> Subject: Re: Historical events
>>>>
>>>> I think that ages are qualitatively different from events. Consider a very short age - 19 June 2018. As far as I am concerned that is not an event.
>>>>
>>>>
>>>> peter
>>>>
>>>>
>>>>
>>>> On 06/19/2018 05:06 PM, Anthony Moretti wrote:
>>>> > Look, I'm no temporal expert or anything, but maybe the problem is in
>>>> > English it could be a continuum from "event" to "period" as the
>>>> > interval of time being named gets longer, with no clear boundary.
>>>> >
>>>> > Mike's birthday party - an event
>>>> > The Middle Ages - a period
>>>> >
>>>> > Because from a data modeling point of view they're the same (at least
>>>> > as far as my modeling knowledge goes).
>>>> >
>>>> > So an unpopular solution I'm guessing would be to rename the type to
>>>> > *EventOrPeriod*.
>>>> >
>>>> > So your previous example:
>>>> >
>>>> > The Black Death
>>>> > superEventOrPeriod: The Middle Ages
>>>> >
>>>> > People's ideas?
>>>> >
>>>> > Anthony
>>>> >
>>>> > On Tue, Jun 19, 2018 at 4:10 PM Richard Wallis
>>>> > <richard.wallis@dataliberate.com
>>>> > <mailto:richard.wallis@dataliberate.com>>
>>>> > wrote:
>>>> >
>>>> > Anthony,
>>>> >
>>>> > I’m not following your logic here. I don’t see a Period (of time from a
>>>> > start time/date to an end date/time) as an event.
>>>> >
>>>> > ~Richard.
>>>> >
>>>> > Richard Wallis
>>>> > Founder, Data Liberate
>>>> > http://dataliberate.com
>>>> > Linkedin: http://www.linkedin.com/in/richardwallis
>>>> > Twitter: @rjw
>>>> >
>>>> > On 20 June 2018 at 00:05, Anthony Moretti <anthony.moretti@gmail.com
>>>> > <mailto:anthony.moretti@gmail.com>> wrote:
>>>> >
>>>> > That definitely works, but only if some definition of Period was
>>>> > agreed on.
>>>> >
>>>> > The property periodEventOccurred would be a subproperty of
>>>> > superEvent in any case:
>>>> >
>>>> > superEvent
>>>> >
>>>> > periodEventOccurred
>>>> >
>>>> >
>>>> > So you could describe the same information using the existing term
>>>> > right?
>>>> >
>>>> > The Black Death
>>>> >
>>>> > superEvent: The Middle Ages
>>>> >
>>>> >
>>>> > Anthony
>>>> >
>>>> >
>>>> > On Tue, Jun 19, 2018 at 3:50 PM Richard Wallis
>>>> > <richard.wallis@dataliberate.com
>>>> > <mailto:richard.wallis@dataliberate.com>> wrote:
>>>> >
>>>> > Using a Period Type I would suggest a cleaner alternative way of
>>>> > saying an event occurred during a period would be like this:
>>>> >
>>>> > {
>>>> > "@context": "http://schema.org",
>>>> > "@type": "Event",
>>>> > "name": "The Black Death",
>>>> > "Description": "A pandemic that spread throughout Europe",
>>>> > "periodEventOccured": {
>>>> > "@type": "Period",
>>>> > "name": "The Middle Ages",
>>>> > "approximateStartDate": "400AD",
>>>> > "approximateEndDate":"1500AD"
>>>> > }
>>>> > }
>>>> >
>>>> > On 19 June 2018 at 23:41, Anthony Moretti
>>>> > <anthony.moretti@gmail.com <mailto:anthony.moretti@gmail.com>>
>>>> > wrote:
>>>> >
>>>> > I was referencing the development version of Schema, I
>>>> > should probably reference production, sorry Roger:
>>>> >
>>>> > * https://schema.org/subEvent
>>>> > * https://schema.org/superEvent
>>>> >
>>>> > To say some event happened during the Iron Age for example:
>>>> >
>>>> > Invention of iron plow
>>>> > superEvent: Iron Age
>>>> >
>>>> >
>>>> > Anthony
>>>> >
>>>> > On Tue, Jun 19, 2018 at 3:05 PM Muri, Allison
>>>> > <allison.muri@usask.ca <mailto:allison.muri@usask.ca>> wrote:
>>>> >
>>>> > The google Cloud host is really slow and I think the 404
>>>> > is a result of something loading too slowly. I could
>>>> > probably publish this more reliably on my own website! I
>>>> > generally just wait a bit and reload the page. Sorry
>>>> > about that.
>>>> >
>>>> > Sent from my iPhone
>>>> >
>>>> > On Jun 19, 2018, at 3:58 PM, Roger Rohrbach
>>>> > <roger@ecstatic.com <mailto:roger@ecstatic.com>> wrote:
>>>> >
>>>> >> I get 404 Not Found for both of those pages.
>>>> >>
>>>> >>
>>>> >>> On Jun 19, 2018, at 11:50 AM, Anthony Moretti
>>>> >>> <anthony.moretti@gmail.com
>>>> >>> <mailto:anthony.moretti@gmail.com>> wrote:
>>>> >>>
>>>> >>> Isn't it already modeled by these properties?:
>>>> >>>
>>>> >>> * https://webschemas.org/subEvent
>>>> >>> * https://webschemas.org/superEvent
>>>> >>>
>>>> >>> Events can exist in part-whole hierarchies, aren't
>>>> >>> named periods just events high in these hierarchies?
>>>> >>>
>>>> >>> Anthony
>>>> >>
>>>> >
>>>> >
>>>>
>>>>
>
Received on Wednesday, 20 June 2018 10:30:03 UTC