Re: Starting the Traffic Event CG

Hi,

After some iterations, I decided not to use PROV ontology because it didn't
fit too well to the traffic model. Instead, after some research I found
some examples with LODE: linked events ontology
http://linkedevents.org/ontology/. Although this ontology it's designed to
be very simple, it brings a lot of links between well-known ontologies,
like DUL, Dublin Core, GEO, etc., therefore is very easy to reuse the most
typical concepts like dates, locations, etc.

This version of the ontology follows the guidelines set in my previous
emails, including a little traffic event classification, vehicle and person
modelling, etc. Please find attached the last version of the ontology, some
RDF examples (one abnormalTraffic and one Accident1) build programmatically
using the Jena Framework in Java.

Also, I have been testing the features of the ontology using SPARQL
queries. As example, this query gives as result a list of towns (results
mashed up with dbpedia.org) with more than 2000 inhabitants which are near
to the coordinates of the points of the itinerary of the traffic event.

PREFIX geo: <http://www.w3.org/2003/01/geo/wgs84_pos#>
PREFIX tff: <http://www.w3.org/community/traffic/>
PREFIX list: <http://www.co-ode.org/ontologies/list.owl#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX dul: <http://www.loa-cnr.it/ontologies/DUL.owl#>
PREFIX dbo: <http://dbpedia.org/ontology/>
PREFIX dbp: <http://dbpedia.org/property/>
PREFIX dbr: <http://dbpedia.org/resource/>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

SELECT ?id ?itinerary ?point ?townName ?population FROM
<abnormalTraffic.rdf>
WHERE
{
?point geo:latitude ?latitude .
?point geo:longitude ?longitude .
?itinerary tff:hasPoints ?point_init .
?listItem rdf:type tff:SpatialThingList .
?listItem list:hasContents ?point .
?id dul:hasLocation ?itinerary
 SERVICE <http://dbpedia.org/sparql> {
?s dbp:settlementType dbr:Municipalities_of_Spain .
?s geo:lat ?dbolat .
?s geo:long ?dbolong .
?s rdfs:label ?townName .
OPTIONAL {?s dbo:populationTotal ?population}
 FILTER (?dbolong > ?longitude - 0.1 && ?dbolong < ?longitude + 0.1 &&
?dbolat > ?latitude - 0.1 && ?dbolat < ?latitude + 0.1)
FILTER (?population > 2000)
FILTER (lang(?townName) = 'es')
}
}

[image: Imágenes integradas 1]

It has been tested using Jena 2.11.1 (with the command-line ARQ).

I believe there are a lot of possibilities opened now. From my point of
view, I think the best option would be review and finish the version of
this basic traffic ontology and, after that, begin to documentate it in the
website.

Regards,
David



2014-06-10 12:00 GMT+02:00 David Torres <datoga@gmail.com>:

> - Person: it is defined in PROV model like "Agent", who does an "Activity"
> that affects to an "Entity". Maybe this concept should be specialized to
> "Driver", "Passenger", "Pedestrian"..... What is SSID? Security insurance
> ID?
>
> - Vehicle: DATEX II has a widely studied classification of vehicles and
> their features (by the way, I'm still waiting to response about the enquiry
> about IPR of DATEX II).
>
> - Weather condition. In this case I would separate this two situations:
>     * General weather conditions (and non weather, low sun glare, fire are
> definitely non-weather problems). This should be defined in a external
> "weather" ontology. Unfortunately I've been unable to find a standard
> ontology, only local extended efforts:
> http://www.semantic-web-journal.net/sites/default/files/swj281_0.pdf.
>    * Road state conditions (example, icy patches): this is directly
> related to a road so it should be defined.
>
> - Organization: PROV model kind of "Agent".
> - Anything: PROV model kind of "Agent" and "Entity".
> - Health stuff: it could be linked to geospecies taxonomy and maybe health
> ontologies to include this information, although after a little review, the
> best could be to include some enumerations and use it directly.
>
> Regarding the Road Accidents I believe these definitions and other (type
> and preliminary causes of the accident) should be also included when the
> concept would be developed.
>
>
>
>
> 2014-06-05 18:18 GMT+02:00 Daniel Dardailler <danield@w3.org>:
>
> So, following up on what I said, I think our ontology should also consider
>> that in addition to Event, a generic description of
>>  - Person (e.g. with an age, SSID, address, desc, etc)
>>  - Vehicle (with insurance, ID, brand, color, type, type of licence to
>> drive them, potential damages, etc)
>>  - WeatherCondition (rain, snow, ice, night, light, etc)
>>  - Organization (e.g Insurance, driver licence's bureau, etc)
>>  - Anything (e.g. Train, Electric Line, Rock)
>>  - HealthStuff (e.g. animal, human, medical condition, types of injuries,
>> drug conditions, etc)
>>
>> exist somewhere, is not specific to traffic, and is not really our job,
>> or is secondary (we need to check if they exist and to what extend they
>> have what we need)
>>
>> In the case of RoadAccident, as a subclass of TrafficEvent, we should
>> focus on
>>  - definition of driver, role, passenger, pedestrian
>>  - defitinition of roadtype, roadconditions, lane#, violations
>>  - relation between vehicles, passenger, driver, pedestrian, moving,
>> parker, etc
>>  - description of cause, context, speed, drugcondition, presence of
>> protection, etc
>>
>>
>>
>>
>>
>>
>> On 2014-06-05 17:42, Daniel Dardailler wrote:
>>
>>> I'm not an expert either, I think at some point we may want to ask an
>>> owl expert through other W3C list (e.g. for the need of constructs
>>> like Comment, ID, hasVersion, or hasSomething in general as a way of
>>> organising semantics).
>>>
>>> At a higher level of design, looking at your base ontology, and in a
>>> good spirit of reuse (the opposite of the NIH spirit, "not invented
>>> here", which I had to fight all my life as a standardizer ;), I wonder
>>> if basic location stuff like
>>>   - point (where you reuse lat/long from geo:)
>>>   - area, segment
>>>   - neartown, roadkm, locationname, etc
>>>
>>> shouldn't be taken from the base geo stuff, where it's got to be IMO.
>>> Or asked if they have plan to have it there. Or if it's OK that we do
>>> it in Traffic.
>>>
>>> TrafficLocation could be a subclass of this base (but powerful)
>>> location and add primary, secondary, etc. maybe roadnumber if that's
>>> not generic enough for the geo ontology.
>>>
>>> Same thing for Event, I'd add a Date BTW, but I'd bet there is a
>>> generic definition somewhere with:
>>>   - date
>>>   - location (as a compound object with point, road, nearby, etc)
>>>   - desc (string)
>>>
>>> and a TrafficEvent could add in its subclass things like
>>> datepublished, dateRange to store things like roadwork for 12h, or
>>> have startEvent and endEvent as subclasses.
>>>
>>> So in short, how about assuming that the really generic classes like
>>>   Location (with rich semantics already, e.g. used in gmap)
>>>   Date  (as used in email)
>>>   Event (simple enough we can define it: Date, Location, Desc)
>>>
>>> will come from some external ontology, not traffic ?
>>>
>>>
>>>
>>>
>>>
>>> On 2014-06-04 17:27, David Torres wrote:
>>>
>>>> Hi,
>>>>
>>>> After some experiments, I have developed a tiny ontology about traffic
>>>> events (please find attached the OWL/XML result and a small graph
>>>> about the ontology). I have used Protégé 4.3 (last stable version)
>>>> to design the hierarchy and include some properties, some of them
>>>> pointing to other ontologies (e.g. basic geo
>>>> ontology, http://www.w3.org/2003/01/geo/ [6] or
>>>> dbpedia http://dbpedia.org/ [7]). I believe this may bring powerful
>>>> features based on the Open Linked Data characteristics.
>>>>
>>>> I'm not an expert in OWL and RDF technical details (probably the
>>>> ontology has something wrong), so it would be great if a real expert
>>>> would take this OWL and make some comments about it!
>>>>
>>>> Regards,
>>>> David
>>>>
>>>> 2014-06-03 16:40 GMT+02:00 David Torres <datoga@gmail.com>:
>>>>
>>>>  Hi, see below
>>>>>
>>>>> 2014-06-02 23:16 GMT+02:00 Daniel Dardailler <danield@w3.org>:
>>>>>
>>>>> Hello David, all
>>>>>
>>>>> see below
>>>>>
>>>>> On 2014-06-02 11:15, David Torres wrote:
>>>>> Hello Daniel,
>>>>>
>>>>> I will try to give a short description about the organisation of
>>>>> all
>>>>> matters related to DATEX II. Easyway was a european research
>>>>> project
>>>>> (I said "was" because it ended some months ago and now they are
>>>>> reorganising their groups) promoted by the EC. Their main task was
>>>>> to
>>>>> develop specs (including methodology, models, documents, xsd,
>>>>> applications, examples, etc.) and provide these specs to the WG8,
>>>>> dependent group of CEN (TC278, ITS standardisation) to be
>>>>> normalised
>>>>> as Technical Specification.
>>>>>
>>>>> The current roadmap of normalisation is:
>>>>>
>>>>> - Part 1: Modelling methodology: normalised (2011).
>>>>> - Part 2: Location referencing: normalised (2011)
>>>>> - Part 3: Traffic information: normalised (2011).
>>>>> - Part 4: Variable Message Settings: normalised (2013)
>>>>> - Part 5: Measured and Elaborated Data: normalised (2014).
>>>>> - Part 6: Parking information: ongoing (probably 2015).
>>>>>
>>>>>  Regarding your concerns about the IRP, of course it must be
>>>>> considered. As far as I know, the only process you have to fulfill
>>>>> in
>>>>> order to work with DATEX II is to download and use it. Anyway, I
>>>>> still
>>>>> believe a formal request for permission is always a good practice.
>>>>>
>>>>> There are (at least) two different things to consider. First there
>>>>> is the permission to point at and download the DATEX spec (for free,
>>>>> usually not the case for CEN/ISO, etc, but seems fine here), so to
>>>>> use it, e.g. as a reference or to implement something based on it,
>>>>> without having to pay some royalty to the IPR owners of the
>>>>> semantics behind the names in the spec (to be checked, but being
>>>>> funded through EC funding, it's got to be RF).
>>>>>
>>>>> Then there is the right to do derivative work, that is, to take
>>>>> pieces and bits, e.g. an enum here, a type construct there, and to
>>>>> define something new, which doesn't need the original work anymore.
>>>>> Usually it's not ok to do that for de-jure spec like from ISO, or an
>>>>> ESO like CEN. In a CG, we do pre-standardization, so it's usually OK
>>>>> to experiment, but if we expect a potential take up in a "real" W3C
>>>>> working group later on, we better check this one.
>>>>>
>>>>
>>>> Ok, I have some contacts in CEN and working groups so definitely I
>>>> will contact them in order to avoid future issues.
>>>>
>>>>
>>>>
>>>>  With "good starting point" I meant it could be taken as
>>>>>> reference. The
>>>>>>
>>>>>
>>>>> Is it even possible to link to a DATEX type or enum if we plan to
>>>>> reformalize their tree schema model in terms of graph/ontologies ?
>>>>> We're bound to cut&paste the strings no ? Could we just point at the
>>>>> definitions then ?
>>>>>
>>>>
>>>> Technically, I'm not sure it's possible to make links directly. The
>>>> main model is exported to a XSD (last version:
>>>> http://www.datex2.eu/sites/www.datex2.eu/files/DATEXIISchema_2_2_0.xsd
>>>> [8]), but this XSD has not URIs or other semantic information so maybe
>>>> it'd need some work to be more "usable".
>>>>
>>>>
>>>>
>>>>  scope... I think it has to be defined. As long as the group is
>>>>>> named
>>>>>> "traffic ontology", I suppose the best option should be to define
>>>>>> the
>>>>>> domain of traffic widely and not to constraint to a one only
>>>>>> subfield.
>>>>>> My proposal would be to use a top-down approach, with these
>>>>>> phases:
>>>>>>
>>>>>> - Define a general traffic domain, including a minimal set of
>>>>>> information about a traffic incident: ID, Date, General type,
>>>>>> location, urgency and level of service. This information is more
>>>>>> or
>>>>>> less the same information that RDS-TMC (ISO 14189) supports (one
>>>>>> of
>>>>>> the main objectives of RDS-TMC was precisely to deliver the
>>>>>> necessary
>>>>>> information to users to be aware of a traffic situation in their
>>>>>> route).
>>>>>>
>>>>>
>>>>> Is this specification freely available ?
>>>>>
>>>>
>>>> No, it's not. Unlike DATEX II the specification is not open, but, due
>>>> to is a very extended technology, its functional details are well
>>>> known. http://en.wikipedia.org/wiki/Traffic_message_channel [9]
>>>>
>>>>
>>>>
>>>>  I'd really like to see an example of definition of a generic traffic
>>>>> event using on ontology language of some kind, e.g. OWL, and
>>>>> referencing this ISO work, to understand what
>>>>> dependencies/extensions are present.
>>>>>
>>>>
>>>> I can prepare something in the following days...
>>>>
>>>>  - Describe a list of traffic subdomains (Accident, Roadworks,
>>>>>> Management, etc.) and a set of reusable domains (Vehicle
>>>>>> classifications, data quality description, etc.), which could be
>>>>>> used
>>>>>> for the rest.
>>>>>>
>>>>>
>>>>> Sounds good. We can use a wiki to share our ideas. I just created a
>>>>> page for our CG:
>>>>> https://www.w3.org/community/traffic/wiki/Main_Page#Draft_Plan [1]
>>>>> Feel free to add there, remove, or create new pages, etc.
>>>>> You should have the rights for editing.
>>>>>
>>>>
>>>> Perfect, I've just added some tasks and possible deliverables.
>>>>
>>>>
>>>>
>>>>  - In several subgroups, work in each subdomain separately,
>>>>>> relating it
>>>>>> to the main traffic domain.
>>>>>>
>>>>>
>>>>> I suppose we'll have to prioritize the subgroup deliverables, as
>>>>> it's likely we'll have to work in series rather than in parallel
>>>>> (unless lots of volunteers show up suddenly)
>>>>>
>>>>
>>>> Indeed!
>>>>
>>>>
>>>>
>>>>  This top-down approach would have the advantage that you get a
>>>>>> final
>>>>>> version very early. This allows to start to build frameworks,
>>>>>> application tools, etc. For example, it could be possible to set
>>>>>> a
>>>>>> complete broadcasting system, from data sources to users.
>>>>>>
>>>>>
>>>>> Does broadcasting (e.g. using the FM band) bring requirements on
>>>>> performance ? size, etc ?
>>>>>
>>>>
>>>> I see broadcasting applications using Internet and avoiding to use FM,
>>>> DAB or other more restricted technologies. Nowadays, a vast majority
>>>> of people are connected when they are driving (connected cars are
>>>> coming but, meanwhile, to have a smartphone with Internet access
>>>> inside a car is not a big deal).
>>>>
>>>> In this case, always following some limitations (e.g. please don't
>>>> send real-time CCTV stream using this kind of channel!), the
>>>> performance or size is not a problem. I think currently it's possible
>>>> to process and inference RDF in a tiny client in an app installed in
>>>> your Android / IPhone.
>>>>
>>>>  About deliverables, examining the websites of other well-know
>>>>>> ontologies, there are always: textual documentation, modelling
>>>>>> graphs
>>>>>> (UML, graph build with Protégé, etc.), resources (including the
>>>>>> ontology, using OWL), use cases and examples). I'm comfortable
>>>>>> with
>>>>>> UML in general, but tools like Protégé have been designed to
>>>>>> develop
>>>>>> ontologies and they have some interesting tools integrated
>>>>>> (graphical
>>>>>> builder, reasoner, OWL profile calculator, etc.).
>>>>>>
>>>>>
>>>>> I'll use whatever tool I can install on Windows.
>>>>> Best would be a webapp tools though (e.g. webprotegé)
>>>>>
>>>>
>>>> We'll give it a try.
>>>>
>>>>
>>>>
>>>>  Finally, I have no problem being volunteer performing, for
>>>>>> example,
>>>>>> some steering or technical tasks. Always with the limitation of
>>>>>> the
>>>>>> time, a few hours per week is ok for me, as you pointed.
>>>>>>
>>>>>
>>>>> Thanks for that.
>>>>>
>>>>> I think we should start with a teleconf, or just a call with a
>>>>> couple of other potential lead/chairs for the group.
>>>>>
>>>>> Last time I attempted to organize a call, I failed finding a proper
>>>>> teleconferencing tool usable by all participants.
>>>>> (we have one at W3C, but not available to CG/the general public)
>>>>>
>>>>> But we can start by designing datatypes through the wiki too, it's
>>>>> more fun ;)
>>>>>
>>>>
>>>> Yes, the time change is always a problem to arrange this kind of
>>>> meeting, maybe the best option is to start to work and wait to get
>>>> feedback from the other people of the group.
>>>>
>>>> David
>>>>
>>>>
>>>>
>>>>  I hope I have answered all your questions.
>>>>>
>>>>> Regards,
>>>>> David
>>>>>
>>>>> 2014-05-31 13:46 GMT+02:00 Daniel Dardailler <danield@w3.org>:
>>>>>
>>>>> Hello David
>>>>>
>>>>> Thanks for the new pointers, it's important to be knowledgeable of
>>>>> this work which seems well advanced and implemented. I need to
>>>>> better understand what are CEN's and the EC roles in these
>>>>> standards, as there may be political aspects to be considered. IPR
>>>>> is one potential source of concerns when someone starts a new
>>>>> standard from some existing standard.
>>>>>
>>>>> A few comments.
>>>>>
>>>>> First, on the logistics side, as I explained before, what this CG
>>>>> (Community Group) needs is a couple of volunteers willing to act as
>>>>> co-chair, editors, organizing teleconf, tracking progress,
>>>>> reporting, etc. That is, ready to spend a few hours per week on it
>>>>> at a minimum.
>>>>>
>>>>> If some folks are willing to try, I'm willing to spend some time
>>>>> explaining how this group mgnt is done at W3C, e.g. live on the
>>>>> phone.
>>>>>
>>>>> On the technical side, I need to look more closely at the DATEX II
>>>>> XML schema
>>>>>    http://www.datex2.eu/content/xml-schema [2] [4]
>>>>>
>>>>> but I already see it includes a full descriptor for Accident, with
>>>>> Vehicle, person, roadtype, etc. And that it's only a tiny portion
>>>>> of
>>>>> the whole semantics provided by DATEX.
>>>>>
>>>>> What do you have in mind when you say "good starting point" ? Only
>>>>> the Accident part ?
>>>>>
>>>>> What would be your ideal deliverable(s) for this CG BTW ? Both in
>>>>> terms of scope (just accident, or traffic in general) and choice of
>>>>> design formalism (tree, graph, oo, xml, etc).
>>>>>
>>>>> I've been pushing for describing Web ontologies in RDF/OWL or some
>>>>> equivalent high level graph oriented framework from the start, but
>>>>> some people were happy with XML as a starting point for instance.
>>>>> There is no obligation from the W3C point of view to choose one or
>>>>> the other. My concerns with starting with XML is that in practice,
>>>>> XML users tend to recreate their own adhoc framework to simulate
>>>>> multiple inheritance, or other logical or graph-like features
>>>>> already present (and standardized) in a richer language (such as
>>>>> OWL). My only constraint is that I don't want to have to learn this
>>>>> sort of XML-improved layer. On the other hand, I'll happily invest
>>>>> time in learning a graphical-oriented ontology editor kind of tool.
>>>>>
>>>>> In any case, scoping should be our first task. My preference goes
>>>>> to a limited scoping: road accident in their context (geo/date,
>>>>> vehicle, driver condition, road type, etc). And try to think graph
>>>>> out of the box.
>>>>>
>>>>> On 2014-05-29 17:03, David Torres wrote:
>>>>>
>>>>> Hello everyone,
>>>>>
>>>>> My name is David Torres, I work as Technical Consultant at
>>>>> University
>>>>> of Valencia (Spain). There is an emerging interest in Spain on
>>>>> Linked
>>>>> Open Data matters and I was looking for a standard ontology based
>>>>> on
>>>>> Traffic Information. The only group I've found is this one and
>>>>> I've
>>>>> just joined, but it seems is quite "idle"...
>>>>>
>>>>> I'm a kind of expert in DATEX II, a traffic model promoted in
>>>>> Europe
>>>>> to be the standard to use to exchange all kind of information
>>>>> related
>>>>> to traffic data. I think it could be a good starting point. The
>>>>> PSM
>>>>> (Platform Specific Model) is based in XML.
>>>>>
>>>>> Description: DATEX II (ISO 16157)
>>>>> Relevance: all kind of incident, accident, traffic data, parking
>>>>> information, CCTV, VMS, etc. It supports the use of extensions.
>>>>> Link: http://www.datex2.eu/ [3] [1] [1]
>>>>>
>>>>> Date: Current
>>>>> Status: In operation
>>>>> Language: English
>>>>> Dataset: Many (in Europe) http://www.datex2.eu/datex-node [4] [2]
>>>>> [2]
>>>>>
>>>>> PS: As long as I can see this mail list is frozen I'd like to get
>>>>> feedback about your interest by working on this new traffic
>>>>> ontology.
>>>>>
>>>>> Kind regards,
>>>>>
>>>>> David
>>>>>
>>>>> Links:
>>>>> ------
>>>>> [1] http://www.webkb.org/kb/nit/accidentOntology.html [5] [3]
>>>>> [2] http://www.datex2.eu/datex-node [4] [2]
>>>>>
>>>>
>>>>  Links:
>>>>  ------
>>>>  [1] http://www.datex2.eu/ [3]
>>>>  [2] http://www.datex2.eu/datex-node [4]
>>>>  [3] http://www.webkb.org/kb/nit/accidentOntology.html [5]
>>>>  [4] http://www.datex2.eu/content/xml-schema [2]
>>>>
>>>>
>>>>
>>>> Links:
>>>> ------
>>>> [1] https://www.w3.org/community/traffic/wiki/Main_Page#Draft_Plan
>>>> [2] http://www.datex2.eu/content/xml-schema
>>>> [3] http://www.datex2.eu/
>>>> [4] http://www.datex2.eu/datex-node
>>>> [5] http://www.webkb.org/kb/nit/accidentOntology.html
>>>> [6] http://www.w3.org/2003/01/geo/
>>>> [7] http://dbpedia.org/
>>>> [8] http://www.datex2.eu/sites/www.datex2.eu/files/
>>>> DATEXIISchema_2_2_0.xsd
>>>> [9] http://en.wikipedia.org/wiki/Traffic_message_channel
>>>>
>>>
>>
>