ICAN Controlled Vocabularies Work in Progress
ICAN Controlled Vocabularies Work in Progress
Page last updated, 7:36 p.m. Pacific time, 6 Feb 2008 | Return to ICAN Working Groups Page
Minutes of Tech Meetings
CMRC GeoNetwork testing site
Luis' ontology with super terms
Latest MIDA theme ontologyNew!
MIDA CV Keywords Notes file (pdf)
The group is using Protégé to build controlled vocabularies. Choose v. 3.2.1.
Project for Temporal keywords:
Temporal.owl
Temporal.pprj
Temporal.repository
Temporal_Jambalaya.properties
Another example
Suggested convention for URIs:
http://xxxx.org/ont/20070823/places.owl
In BNF:
::= "http://" "/" "/" "/" ".owl" := | := | := “.” ::= ( [ “_” ] ) |
February 6, 2008 from Tanya
Hi Tech folks,
Had a thought about the need to refine the global ontology we are using, and thought it might be worth suggesting via email (even though we don't have a conference call scheduled yet):
Since we simply lifted the top terms from the existing USGS controlled vocabulary on coastal hazards, what do you think of now adopting the next level (all or in part) from their vocabulary in order to narrow the our terms further? This possible shortcut would give us a good start in getting to a set of finer global keywords.
for example:
Agents_of_Coastal_Change.Natural_Processes
Agents_of_Coastal_Change.Human_Activities
Agents_of_Coastal_Change.Sediment_Budget
Effects_of_Coastal_Change.Shoreline_Erosion
Effects_of_Coastal_Change.Shoreline_Accretion
Effects_of_Coastal_Change.Shoaling
Effects_of_Coastal_Change.Flooding
Effects_of_Coastal_Change.Habitat_Alteration
Human_Responses_to_Coastal_Change.Emergency_Response
Human_Responses_to_Coastal_Change.Disaster_Recovery_and_Reconstruction
Human_Responses_to_Coastal_Change.Legislation
Human_Responses_to_Coastal_Change.Mitigation_Strategies
Human_Responses_to_Coastal_Change.Policy
Human_Responses_to_Coastal_Change.Preparedness
etc.
To see the full vocabulary list they use see here
Perhaps both OCA and MIDA can look at the USGS list and see if the second level contains any terms that are not appropriate, and / or suggest terms to add that might cover scenarios not on the USGS list. From there we should be able to emerge with a second tier to our global ontology.
What do you guys think? It seems worth us each examining this option internally before we have a conference call, so that we can be prepared to discuss either acting in this direction or another.
February 6, 2008 from Tanya
Thanks Declan, I'll apply the patch. I've been using Firefox so would not have noticed this if you hadn't pointed it out.
A few other things to report:
1) Our GeoNetwork 2.1.0 is installed with a few metadata records and seems to be behaving as expected. I'll forward the URL soon.
2) I think I mentioned this back in November sometime, but I've noticed some bugginess to do with importing certain kinds of metadata. MIDA may not be affected by this problem because if you remember our metadata is in the FGDC standard, and I suspect this issue may only apply to that standard.
Now that I've spent more time looking into this, I know what to avoid and I don't seem to be having any problems. However the problematic metadata seemed to be the ESRI formatted FGDC XML, so I thought I would mention it for the record.
Luckily, since we already manage our metadata in a database (SMMS), the FGDC XML records exported from the SMMS database go nicely into the GeoNetwork, with no complaints. I haven't examined the ESRI XML in too much detail, but I think that there may be some vendor-specific tags in there that are causing GeoNetwork to report errors on import. When errors are reported, it is not clear (from the error message) if import has failed or not, and manual confirmation by checking the database is required. I'll leave this discussion here for the moment, as I am doubting it affects MIDA, but when I get around to it, I'm going to try to come up with a solid example to report to the GeoNetwork list, and see if it is a wider problem with a solution or not.
3) Ontology Status and public URLs:
Once I have posted an updated version of the ontology, I will post that URL and also the GeoNetwork URL. One thing I did want to ask is if it is ok for me to leave the GeoNetwork on port 8080 ? will that cause any problems for the prototype?
February 6, 2008 from Declan
There is a small bug in GeoNetwork 2.1.0 that affects IE7 browsers. This
relates to visualizing metadata within the web GUI for GeoNetwork.
However, this does not affect ICAN demo as we mainly use CSW server part
of GeoNetwork. But still, the available bug fix for this problem is
useful for internal visualization of metadata, whether internally for
MIDA or OCA, etc.
Attached is a fix for this problem. You need to copy the attached file (gn_search.js_bugfix.txt) to .../web/geonetwork/scripts/gn_search.js
For the record, the bug is described in this thread
Also, GeoNetwork 2.2.0 is under development and currently a release
candidate since January 08. So this should include more bug fixes.
However, for the moment we should stick to GeoNetwork 2.1.0 and use this
for first version of ICAN demo. After first version of ICAN is
successful, and when GeoNetwork 2.2.0 is fully released we can
investigate if upgrading to GeoNetwork 2.2.0 is beneficial. So we should
not spend many days of work putting 100's of metadata records into
GeoNetwork (just example metadata records at the moment) until ICAN
demo, etc is stable.
November 12, 2007 from Liz
I've put the latest MIDA ontology online (mida.ucc.ie/ont/20071109/theme.owl). We have edited down our controlled vocabulary to include only terms for data that we have or could fairly easily acquire. We then mapped those terms to the Superterms (these can be seen in the online version by going to mida.ucc.ie/ont/20071109/theme and then scroll down to the terms that begin with "super:"). I'm still waiting for it to be reviewed by our coastal geomorphologist so there will be another version beyond this, but I wanted to get it out before our conference call tomorrow.
October 30, 2007
October 19, 2007 from Luis
1) Looking at your XML and the XML sent by Declan it appears that we are looking at different ISO 19139 Schemas.
2) I found the TC211 schemas available here: www.isotc211.org/2005/gmd
3) I wrote 2 proposals - maybe as a starting point about how to
encode URIs in CSW. Other groups are also interested in this issue,
so I put the documentation in the MMI web site (marinemetadata.org/uriiso).
Please comment ...
October 19, 2007 from Declan
An example CSW response linked to a WMS, requested after Wednesdays tech conference call.
I have installed latest version of Geonetwork on my desktop computer. I will install
the latest version of Geonetwork on our main server by next tech conference call.
So, I cannot give live links. Instead, I have attached CSW responds. Only the
SUMMARY and FULL CSW responses contain the WMS info.
***** linked files *********
iso19139.xml:
A quick test iso19139 xml record that I created through Geonetwork metadata
entry tool
CSW_getRecords_response_isoRecord_brief.xml:
Response from CSW getRecords asking for a BRIEF iso Record
CSW_getRecords_response_isoRecord_summary.xml:
Response from CSW getRecords asking for a SUMMARY iso Record
CSW_getRecords_response_isoRecord_full.xml:
Response from CSW getRecords asking for a FULL iso Record
*********************
Note: Only SUMMARY and FULL return WMS information
the URL stub to the CMRC WMS server is “http://interrisk.ucc.ie/ogc.php?”
the WMS layer id layer is “coastGshhs” (an Irish/UK coastline layer example)
***** wms info within metadata *********
.
.
.
.
.
.
.
.
.
*********************
The main WMS operations are GetCapabilities, GetMap, GetLegendGraphic and GetFeatureInfo. For the ICAN demo, we will use the GetMap operation to visualise a map.
** GetMap **
We can construct a GetMap query string from this “http://interrisk.ucc.ie/ogc.php?” and “coastGshhs” info.
URL format…
* http://interrisk.ucc.ie/ogc.php?
SERVICE=WMS&
REQUEST=GetMap&
VERSION=1.1.1&
BBOX=0,60,10,70& (some bounding box)
LAYERS= coastGshhs& (layer id, can be multiple layers listed here, comma
separated)
SRS=EPSG:4326& (projection)
FORMAT=png& (image format)
WIDTH=600& (image width)
HEIGHT=428& (image height)
Live example to CMRC WMS server (for layer id “coastGshhs”)
(paste into browser if direct does not work)
http://interrisk.ucc.ie/ogc.php?LAYERS=coastGshhs&REQUEST=GetMap&SERVICE=WMS&FORMAT=image%2Fpng&
STYLES=&HEIGHT=428&VERSION=1.1.1&SRS=EPSG:4326&WIDTH=600&BBOX=%2
D12%2E9098360655738%2C50%2C%2D3%2E09016393442623%2C57&
TRANSPARENT=TRUE
*********************
Some extra info…, can explain in more detail later as we look into setting up the WMS servers and designing the portal…
But all this is easy to configure via the UMN Mapserver mapfile, so don’t worry about the detail too much.
** GetCapabilities **
We will NOT need to use this GetCapabilities document for ICAN, if we agree to setup our WMS servers in a certain way…
e.g. support PNG images, support at least WGS84 projection (EPSG:4326), etc.
Live example to CMRC WMS server…
http://interrisk.ucc.ie/ogc.php?SERVICE=WMS&VERSION=1.1.1&REQUEST=GetCapabilities
This gives info such as
* List of all data layers held on this WMS server
* The Bounding Box extent of each individual data layer
* Projections supported by a layer
* Image formats supported by a layer
* Simple metadata like keywords of each individual data layer (optional)
* URL to a graphic legend for an individual data layer (optional)
* etc.
** GetLegendGraphic **
Live example to CMRC WMS server…
http://interrisk.ucc.ie/ogc.php?SERVICE=WMS&VERSION=1.1.1&REQUEST=GetLegendGraphic&LAYER=coastGshhs
&format=image/png
Returns a graphic legend image for layer coastGshhs
October 17, 2007 from Tanya
We have (finally!) put the latest theme cv/ontology on our server, in the same path as designated in the URI (using the structure that Luis recommended). You can see our latest versions of the Oregon Theme OWL at:
www.coastalatlas.net/ont/20071016/theme.owl
As before, you can browse the OWL in web page format by removing the ".owl" from the above link and hitting Return.
Obviously the above is a draft, and probably incomplete or riddled with errors. Not sure if I have the right idea here with how I have gone about linking some of these, but I guess that's what this figuring out process is for. I certainly have a few questions for Luis and Yassine when we get to that stage.
Many thanks to Liz for her helpful instructions on the linking process....
October 8, 2007 from Luis
On 4 October Dr. Tania Tudorache presented a very tidy summary of the state of the art in collaborative ontology development based on Protege. John Graybeal from MMI attended it and said it was very good. I want to share the links with the group. I will review them later this week.
link
presentation (pdf)
audio (mp3)
October 2, 2007 from Tanya
First, our discussion on the 12th resulted in the original idea of 2
super terms:
- Coastal Vulnerability (to Coastal Erosion)
- Human Responses (to Coastal Erosion)
To clarify, it was my thinking that we were generally trying to capture
the physical and human dimensions of coastal erosion risk with these 2
terms. They are very generic, and as Liz points out are maybe too
semantically vague. So below is a summary of the other perspectives
people have suggested, with the idea that we can maybe come up with
something more specific from amongst these if need be.
++++++++++++++++
1. On sept 17th I forwarded an email about hazard risks in a general
sense, which talked about 3 concepts (imagine the phrase "Coastal
Erosion" in front of each of these terms):
- Hazard: assessment of the location, frequency and intensity of
hazard events;
- Vulnerability: assessment of the degree of loss possible due to a
given severity hazard occurrence;
- Exposure: quantification of the spatial distribution and the
estimated value of the potential impacted targets.
The above is a perspective which breaks out our catchall term of
"coastal vulnerability", but which misses the "human response" concept
we wanted.
-------------------
2. The USGS index on coastal vulnerability comprises 6 variables:
- Tidal range
- Wave height
- Coastal slope
- Shoreline erosion rates
- Geomorphology
- Historical rates of relative sea level rise
The USGS index captures the physical/geomorphology aspects of risk
well, but the human aspects not at all.
--------------------
3. The NOAA Coastal services center has a Coastal Risk and
Vulnerability Assessment tool which comprises 6 concepts:
- Hazards Analysis
- Critical Facilities Analysis
- Societal Analysis
- Economic Analysis
- Environmental Analysis
- Mitigation Opportunities Analysis
This NOAA example captures the human aspects of risk well, and is light
on the physical/geomopholgy aspects of risk. The website is also good in
that is suggested necessary data layers for each of these analyses
steps. These lists could be helpful to the MIDA folks for ideas on data
layers. See: www.csc.noaa.gov/rvat/rvat.html
----------------------
4. The Eurosion brochure "Living with Coastal Erosion in Europe" is
focused primarily on cases studies and therefore fairly comprehensively
covers human interactions with coastal erosion. The Introductory section
summarizes the topic using these concepts:
- Causes of Coastal Erosion
- Natural
- Human induced
- Effects of Coastal Erosion
- loss of land (with economical, societal or ecological value)
- destruction of natural sea defenses (potentially leading to
flooding)
- destruction of artificial defenses (potentially leading to
flooding and other loss of assets)
- Mitigating Actions for Coastal Erosion
- Hold the Line
- Do Nothing
- Limited Intervention
- Managed Realignment
From a CZM perspective, the Eurosion document in many ways is most
complete in that it addresses both the physical and human issues. If it
is weak it is on the physical/geomorph side.
+++++++++++++++++++++++++++
That's all for the summary. Hopefully this is helpful in some way for
the discussion tomorrow. Based on this information I think it's likely
we will revise our original decision to have only 2 terms. Perhaps we
could expand to 3 which could encompass the 3 concepts in the Eurosion
outline : Causes, Effects, Mitigating Actions. These are very similar as
it turns out to the upper level terms in the USGS controlled vocabulary:
- Agents_of_Coastal_Change
- Effects_of_Coastal_Change
- Human_Responses_to_Coastal_Change
see mrib.usgs.gov/user/cch/meta/hot_topics.txt for a reminder
on that CV.
October 2 from Liz
Content issues:
Coastal Vulnerability: Talking with Jeremy (our coastal geomorphologist), he's suggested that we change this from vulnerability to sensitivity. This could be another issue of semantics so let me know what you think. His understanding of coastal vulnerability is that it is specific to the physical vulnerability of what can and can't be eroded (geology, sediments etc) - whereas sensitivity would include human factors that can contribute to erosion, such as heavily populated areas, property and impact of use (land use, tourism, shipping channels, etc). Thoughts?
Technical details:
Open up your theme ontology in Protégé. Then you'll need to import Luis's coastal_erosion.owl. Do this by going to the metadata tab and click on the Import ontology… button. Select "import an ontology from the web" and enter the URL of Luis's ontology. You'll be given a list of prefixes. Change the first, p1, to something that identifies the ontology (we named ours "super"). This will help to identify the classes that are imported when you look at the class list (e.g. super:Coastal_vulnerability).
When you are finished you'll see Luis's ontology embedded in yours, in the ontologies list (the text is grey). If you only see the name of your ontology, double-click on it to see the contents. Note that Luis's ontology has another imported ontology (Dublin Core) within it.
In the metadata tab, have a look at the properties that Luis added. I've added similar details to our ontology too, using the "create new annotation value" button.
Then to map the OWL classes with the top ontology, you then go through your list of classes and determine which fall into either of the two categories. When you have a class to map, select it and then look a the "Asserted Conditions" box to the right. Select the "Add named class…" button and then select the appropriate class (i.e. coastal_vulnerability or human_responses) from the top ontology. You'll then see that it appears in the top ontology list.
With the hierarchy of the classes and subclasses, pick the highest class, if all the subclasses apply to the super term. If they don't, you will need to map each of the applicable subclasses and not the higher class. For example, Geology can be mapped to coastal_vulnerability. In MIDA the subclasses include bedrock, seabed, soils and terrestrial. If every subclass can apply to coastal_vulnerability, then we map the class Geology. If we decide that soils don't apply, then we would instead map the subclasses bedrock, seabed and terrestrial to the super term, and do nothing with the class Geology.
An *incomplete* version of our latest theme ontology is available online at mida.ucc.ie/ont/20070929/theme.owl
We'll modify it after discussions on Wednesday.
August 29, 2007
The MIDA group has put the latest controlled vocabularies on their server, in the same path as designated in the URI (using the structure that Luis recommended). You can see our latest versions of the MIDA OWLs (you can save them and open in Protégé as an example) at:
http://mida.ucc.ie/ont/20070829/place.owl
http://mida.ucc.ie/ont/20070829/temporal.owl
http://mida.ucc.ie/ont/20070829/stratum.owl
http://mida.ucc.ie/ont/20070829/discipline.owl
http://mida.ucc.ie/ont/20070829/theme.owl
You can also see them in web page format (not in hierarchical format unfortunately, as you would see it if you opened the OWL in Protégé). To see the pages, remove the ".owl" from the above links and hit Return. Making our ontologies available this way will make them more easily accessible.
These are all works in progress and there will be more modifications in the near future as people internally review them. These mostly reflect MIDA's existing data and some of the potential data. We will be building it up more as we research existing ontologies related to coastal erosion.
To create these web pages for your own CV/ontologies, first create a folder for the current version (e.g., 20070829). Then in that folder create additional folders for the keyword types (place, temporal, stratum, discipline, theme), where their web pages will be created. Next in Protégé go to Tools --> Generate OWLDoc…, then navigate to where you want to save the html files for the relevant keyword type. This will generate a lot of HTML files -- the key is the index.html file in each folder. Then create the appropriate folder on your server (e.g. so it navigates to http://www.coastalatlas.net/ont/
September 7, 2007
You can see our latest versions of the Oregon
OWLs (you can save them and open in Protégé as an example) at:
http://www.coastalatlas.net/ont/20070907/place.owl
http://www.coastalatlas.net/ont/20070907/temporal.owl
http://www.coastalatlas.net/ont/20070907/stratum.owl
http://www.coastalatlas.net/ont/20070907/discipline.owl
http://www.coastalatlas.net/ont/20070907/theme.owl
You can also quickly browse them in web page format by removing the
".owl" from the above links and hitting Return.
These are *seriously* works in progress ... somewhat drafty and have
not yet been reviewed by as many people as I would like, but they are a
good working start for the purposes that we are after at the moment. I
can already think of a few things we may have missed, but those will be
included in future drafts. Definitions are also sparse at the moment,
but I hope to fill them in gradually now that a basic structure is in
place.
Of note: per Yassine's advice, at the moment all concepts are
intentionally represented as classes, pending the evolution from
controlled vocabulary to actual ontology.