RE: On the time model

Hello

There
 can always be fine and coarse grained events. A simple event denoting 
that Claudio moved to Vienna, could be used to create a more complex 
event (or fact?) saying that Claudio lives in Vienna if there is no 
other moves-to
event, at least to the moment of the creation of the complex event 
creation. But I think that doesn't change much in terms of the model. 


However the notion of what is a short event or what is a long-running 
one may differ from use case to use case, so in an abstract model we 
might not impose this.





About the events with unbounded intervals, maybe from the abstract 
modeling point of view we can allow them. But then in practice this can 
be solved by dividing in simpler events. If we want to know 'what 
meetings are currently being held in WebEx', we could
do it with an event denoting that a meeting has started, and if there is
 no corresponding ending event. 


Or we could have a meeting event with an interval that is not closed.


Again, maybe we can have these different possibilites in the model.



Jean-Paul 


From: bernstein@ifi.uzh.ch
Date: Fri, 4 Apr 2014 16:16:53 +0200
CC: public-rsp@w3.org
To: claudio.di.ciccio@wu.ac.at
Subject: Re: On the time model

Hi all
You might want to read the TEF-SPARQL working paper that can be found at:
http://www.merlin.uzh.ch/publication/show/8467

Best
Avi

On 04.04.2014, at 16:13, Claudio Di Ciccio <claudio.di.ciccio@wu.ac.at> wrote:Slight addition to the preceding email, on the possibility of unbounded time intervals.
I would avoid that. If we use the happening-interpretation, it would naturally lead to the need to always define and end for the interval, anyway.


The reason why in my opinion never-ending events should be avoided, is that they should be kept available for new consumers that would possibly connect, forever.



Kind regards,
Claudio

On 4 April 2014 15:56, Claudio Di Ciccio <claudio.di.ciccio@wu.ac.at> wrote:


Hi all.Unfortunately, the microphone seemed to stop working during the Conference Call. Therefore, I was not able to intervene.



This is my major concern about the representation of events, w.r.t. time.
My opinion is that events should report things that happen, not facts. Therefore, the sense of the time interval would change, becoming unrelated to the semantics of the reported fact. In other words, the time interval would not refer to the time range in which the reported fact still holds true, but rather to the time given to event consumers to process the event before it expires.



To clarify my position:1) Events happen, facts remain. Therefore, facts could be considered as infrequently changing.


2) What if we have recurring activities? Say, from November the 11th, e.g., every day from 00:30 to 04:30 the Metro service stops. Is this an event that recurs every day, from Nov 11 onwards, or rather one event occurring on Nov 11 and establishing the rule, valid forever? The fact-reporting-interpretation of events would lead to the second option. However, this would lead to the necessity of defining periodicity for time intervals -- which goes beyond the scope of our group, let alone overcomplicating the model. The happening-oriented approach would allow for the first option and a less complicated model.


3) As far as we decided, we have no alteration of preceding events. Therefore, an event like "Claudio - lives in - Vienna" is forever true, until a new event amends it. However, if a new event comes ("Claudio - lives in - London"), whereas the preceding was without an ending time, we would have an inconsistency: an event contradicts another, and there is no clear way to establish which is saying the truth. Instead, "Claudio - moves to - Vienna" and "Claudio - moves to - London" would make sense, as the start-time and end-time would refer to the time where the event is still processable. After the end-time expires, we might think of some consequences, such as inferring and storing that Claudio lives in Vienna, e.g. This would be related to reasoning, and disregarding the data model.


​Have a nice weekend.​​Best regards,
Claudio​-- 


Dr. Claudio Di Ciccio
WU ViennaInstitute for Information BusinessBuilding D2, Entrance C, 3rd FloorWelthandelsplatz 1, 1020 Vienna, Austria
Email: claudio.di.ciccio@wu.ac.at


Phone: +43 1 31336 5222


-- 
Dr. Claudio Di CiccioWU ViennaInstitute for Information BusinessBuilding D2, Entrance C, 3rd Floor

Welthandelsplatz 1, 1020 Vienna, Austria
Email: claudio.di.ciccio@wu.ac.atPhone: +43 1 31336 5222



-----------------------------------------------------------------
|  Professor Abraham Bernstein, PhD
|  University of Zürich, Department of Informatics
|  web: http://www.ifi.uzh.ch/ddis/bernstein.html