W3C home > Mailing lists > Public > public-vocabs@w3.org > December 2011

Re: Proper schema.org markup of repeating / recurring events?

From: Martin Hepp <martin.hepp@ebusiness-unibw.org>
Date: Wed, 21 Dec 2011 09:14:24 +0100
Cc: public-vocabs@w3.org
Message-Id: <9DABF924-EC10-4B32-8C2A-21D6BB57CB63@ebusiness-unibw.org>
To: Sean Carlos <sean@antezeta.com>

just my two cents: From my experience of designing Web-scale data schemas, it is much better to not hide rules for repeating patterns (e.g. recurring events, train schedules, etc.) in conceptual elements but to materialize all options in the form of data. For example, instead of modeling that a certain event is on every other Wednesday at 2.00 p.m., it is typically better to model each event separately.

The reasons for this modeling practice are:

1. It is much easier to query data containing recurring patterns and single, materialized facts.
For instance, the query for whether there is a Jazz event in town has to check just whether there is such for a given data, instead of backward-chaining all potentially relevant rules for such events.

2. It is impossible to provide constructs for all possible exceptions, e.g. "this event is on every other Wednesday, unless that day is a the birthday of Queen Mum".

3. It is very difficult or impossible to model irregular cases, e.g. if a single concert starts 20 minutes late or is cancelled, or that a single flight is 15 minutes delayed.

But the main reason is that hiding such rule knowledge inside conceptual structures without a formal meaning is bound for trouble and complicates the execution of queries,

It sounds painful at first, but I am deeply convinced that it is better to put the rules knowledge in a simple script that materializes all combinations and then publish these.

If the data is very dynamic (e.g. flight tickets), then provide a query interface that uses the same conceptual elements and tries to answer the query by backward-chaining or translating the query into a call to a proprietary Web service.

For instance, the Tickets Ontology, http://purl.org/tio/ns uses this for modeling all kinds of time-table information, and it works pretty well. While the alignment with schema.org is not yet completed, this seems like a very promising direction.



On Dec 20, 2011, at 4:20 PM, Sean Carlos wrote:

> Hi list,
> I have several cases of repeating / recurring events, such as a
> monthly business networking lunch, where the only data elements which
> change from event to event are the dates.
> It appears to me that schema.org doesn't yet address this particular use case?
> The current solution appears to be to repeat each event, including the
> duplication of each data element.  I'm thinking of using the search
> engine frowned-upon <meta> tags to hide the repeated data in
> successive event dates.
> Insights and alternatives from the trenches would be welcome.
> In a similar vein, the schema.org documentation is silent on how to
> markup multiple events which occur at the same venue.  There is an
> example specific to Google:
> http://support.google.com/webmasters/bin/answer.py?hl=en&answer=164506
> but it would be nice to see this officially documented for schema.org
> AND clarity provided regarding the order of the data - i.e. the
> possibility to list each event occurrence then the venue, i.e. putting
> the event where it belongs, before the venue.
> best,
> Sean Carlos
> -- 
> __________________________________________________
> Antezeta Web Marketing - Search/Social/Email/Analytics
> ITA Tel: +39 02 69 000 333 http://antezeta.it/
> USA Tel: +1 617 394 1100 http://antezeta.com/
> Skype: sean-carlos
> News: http://antezeta.com/news/
>           http://antezeta.it/news/
> Twitter: @sean_it (italiano)
>           @seancarlos (English)
Received on Wednesday, 21 December 2011 08:16:20 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:29:21 UTC