[Fiware-data] FI-WARE-data / VisionDocument / Data-Context-Eventdefinition - a proposed revision of the Event definition

Guy Sharon GUYSH at il.ibm.com
Fri Jan 6 01:36:58 CET 2012


Good idea to start drawing things.
But this is far from what I was trying to convey. In fact, I now 
understand that we need to touch \ sync on a fundamental point about 
events.
Events must be treated as first class citizens in models and transports - 
not incorporated into entities\things and collapsed into properties rather 
they should exist alongside entities and may be associated with entities.
I will try to give an example through three cases - 1) the simple one with 
the boiler, 2) an event to many entities with a check deposit, and 3) many 
events to one entity with the TVSet.
If needed a drawing board and a discussion could be a good idea to resolve 
this in Madrid.

First a simplified ontology view of the three cases
1) 
2)
3)
 

Now lets look at how an event then should be represented as in the Vision 
of the chapter and how they would be stored or communicated versus 
entities.
1)
These may actually be 2 different views on all the data that gets 
collected (repository)
First view through the eye of an entity - you would see all the 
temperature values over time
Second view through the eye of an event - you would see all the 
TemperatureMeasured events over time and what temperature reading they 
provided to what Boiler

2) Here - it is IMPORTANT to keep all the 'properties' of the event 
together (even though they are associated to different entities) for 
proper event processing - this is for some event processing implementation 
a real big issue when they decouple especially when you need to reason 
over the events and not over the entities it may have been associated 
with.


These may actually be 2 different views on all the data that gets 
collected (repository)
First view through the eye of the entities - each entity will show all its 
properties by property over time
Second view through the eye of an event - you would see all the events 
over time with all the relevant 'properties' that are associated to 
different entities - together.

3) Here - it is IMPORTANT to distinguish between different events with 
'properties' that are associated to the same entity property. 


These may actually be 2 different views on all the data that gets 
collected (repository)
First view through the eye of an entity - you would see all the location 
values over time regardless of the type of event
Second view through the eye of an event - you would see all the events 
over time and what location the TVSet is according to the event types.


Regards,

Guy Sharon
Manager
Event-based Middleware & Solutions

Phone: 972-4-8296587 | Mobile: 972-54-6976417
E-mail: GUYSH at il.ibm.com
Website: www.research.ibm.com/haifa/dept/services/soms_ebs.html


Find me on:   and within IBM on:   


Haifa University, Mount Carmel
Haifa, HA 31905
Israel




From:   <fano.ramparany at orange.com>
To:     Guy Sharon/Haifa/IBM at IBMIL
Cc:     <boris.moltchanov at telecomitalia.it>, 
<carlo.licciardi at telecomitalia.it>, <Fiware-data at lists.fi-ware.eu>, 
<jhierro at tid.es>
Date:   05/01/2012 12:13
Subject:        RE: [Fiware-data] FI-WARE-data / VisionDocument / 
Data-Context-Eventdefinition - a proposed revision of the Event definition



Hi Guy,
 
Thank you for your comments. Most of my feedback are covered in the joint 
IoT/Data discussion.
I?d just like to bounce on your attempt to describe the boiler example 
using the data/context/event schema.
 
When you say ?you treat temperature and pressure as attribute?. Is fine 
with me ? but is irrelevant to the above representation?, do you mean that 
you would handle this differently?
In which case could you elaborate?
 
Here is my proposal:
 

Which in the graphical notation corresponds to:
 

 
My point is that it seems more natural to consider the 3 
?DataElementAttribute? boxes in the middle as events,  than to call the 
entire graph an event. In my proposal the links are bidirectional, which 
makes It possible to retrieve the DataElement from the 
DataElementAttribute.
 
Any comment or  alternative viewpoint welcome,
 
Kind regards,
 
Fano
 
 
De : Guy Sharon [mailto:GUYSH at il.ibm.com] 
Envoyé : mardi 3 janvier 2012 20:51
À : RAMPARANY Fano RD-TECH-GRE
Cc : boris.moltchanov at telecomitalia.it; carlo.licciardi at telecomitalia.it; 
Fiware-data at lists.fi-ware.eu; fiware-data-bounces at lists.fi-ware.eu; 
jhierro at tid.es
Objet : Re: [Fiware-data] FI-WARE-data / VisionDocument / 
Data-Context-Eventdefinition - a proposed revision of the Event definition
 
Fano, 

I think the discussion you raise brings up the aspect of the semantic 
meaning of an event and the representation of an event. 
In the definition of an event in our chapter we discuss the representation 
of an event and not the semantic meaning of an event. 
An event may represent change of and attribute value of some entity. It 
can also mean a change in state \ transition. 
But how do you communicate the event of an attribute change for example... 

You would provide an event which is a specific Data Element that consists 
of Data Element attributes that may include info on the entity for an 
attribute is being changed, the attribute that is being change, the old 
value of the attribute, the new value of the attribute, the time during 
which the change occurred, etc. 
Therefore the event representation - how the event is communicated between 
GEs for example is a specific Data Element. 
In your example there are 2 types of events - TemperatureMeasured and 
PressureMeasured 
TemperatureMeasured consists of tempValue, timeStamp and boilerID 
attributes - for example 

I think you refer more to the semantical or ontology level of the event 
which in my opinion is not covered by the whole Data Element discussion. 
>From that point based on the example -  we have an event associated with 
temperature of a boiler and another associated with the pressure of the 
boiler - here you treat temperature and pressure as attributes of the 
boiler and events associated to these attributes - which is fine with me - 
but is irrelevant to the above representation. 
In other words - event element attributes tempValue, boilerID of the event 
element TemperatureMeasured are semantically mapped to the Boiler. 
An event could carry information relevant to more than one entity you wish 
to model which in that case the event does not represent a change in 
attribute but perhaps something more abstract - a situation! 
The output of event processing is also an event - a complex event - that 
may not be associable to an attribute or rather is associated to multiple 
attributes of multiple entities. 

Regards, 

Guy Sharon 
Manager 
Event-based Middleware & Solutions Group 


Event-based Middleware & Solutions 

phone : 
+972 4 8296587 
mobile : 
+972 54 6976417 
address : 
IBM R&D Labs in Israel, Haifa University Campus, Mount Carmel, Haifa, 
31905, Israel 
email : 
guysh at il.ibm.com





From:        fano.ramparany at orange.com 
To:        boris.moltchanov at telecomitalia.it, 
carlo.licciardi at telecomitalia.it, jhierro at tid.es 
Cc:        Fiware-data at lists.fi-ware.eu 
Date:        03/01/2012 20:45 
Subject:        [Fiware-data] FI-WARE-data / VisionDocument / 
Data-Context-Event        definition - a proposed revision of the Event 
definition 
Sent by:        fiware-data-bounces at lists.fi-ware.eu 




Hi, 
I wish you the best for 2012 and in particular much success to our FIWARE 
project and to the FI PPP program in general! 
FT is involved in adding a semantic extension to the Publish/Subscribe GE. 
One step required is to define a ?semantic aware? data schema (*) able to 
capture the Data-Context-Event model that has been defined in the Vision 
Document. 
While revisiting this chapter (the overview chapter: 
https://forge.fi-ware.eu/plugins/mediawiki/wiki/fiware/index.php/Data/Context_Management#Overview
), I felt uncomfortable with the definition of an event, which according 
to the current version is ?specific Data Element? with a timestamp to 
represent its date/time occurence. 
However, events are generally related to the creation of ?element 
attributes? or changes in the values of ?element attributes? (for example, 
if the temperature of the boiler has been measured 2 times and its 
pressure 1 times. There will be 3 events, each of them will correspond to 
a Context Element attribute. More precisely, 2 context attributes for the 
two temperature measurements and 1 context attribute for the pressure 
measurements. The timestamps of those 3 measurements would probably be all 
different because the measurements would be done at different times. The 
?element attribute? corresponding to the 2nd temperature measurement will 
supercede the one corresponding to the 1rst temperature measurement. 
(Note: with the approach proposed in the working document I refer to below 
(**), the timestamp is a meta-data associated to the ?element attribute?). 

For this reason I?d like to propose a slight revision of the event 
definition and state that ?an event is a specific Data Element attribute? 
(with a mandatory timestamp meta data to capture its date/time 
occurrence). This definition would replace this one: ?an event is a 
specific Data Element?. 
Any comment/feedback on this and on the working document which proposes a 
flexible ontology which aligns with our data model, is more than welcome, 
Best regards, 
Fano 
(* called an Ontology in the semantic web framework) 
(** This document proposes an Ontology for Defining Ontologies for the Pu 
ReleaseV0. In case the link doesn?t work, the document is in 
FI-WARE->files->OntologyV0.pdf on the FI-WARE fusion forge web site)
_______________________________________________
Fiware-data mailing list
Fiware-data at lists.fi-ware.eu
http://lists.fi-ware.eu/listinfo/fiware-data
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/gif
Size: 3092 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment.gif>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/gif
Size: 5210 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment-0001.gif>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/gif
Size: 3243 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment-0002.gif>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/gif
Size: 9749 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment-0003.gif>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/gif
Size: 25563 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment-0004.gif>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/gif
Size: 18440 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment-0005.gif>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/jpeg
Size: 518 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment.jpe>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/jpeg
Size: 488 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment-0001.jpe>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/jpeg
Size: 467 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment-0002.jpe>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/jpeg
Size: 494 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment-0003.jpe>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/gif
Size: 360 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment-0006.gif>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/png
Size: 9509 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/jpeg
Size: 11456 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment-0004.jpe>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/gif
Size: 2558 bytes
Desc: not available
URL: <https://lists.fiware.org/private/old-fiware-data/attachments/20120106/5bfae0b7/attachment-0007.gif>


More information about the Old-Fiware-data mailing list

You can get more information about our cookies and privacy policies clicking on the following links: Privacy policy   Cookies policy