Describe CodeValue here.
ontology lives here
TODO:
I second Robyn's comments, Chris Well done .. now let's capitalise on it :o)
John
Original Message
From: Robyn Smits Sent: Wednesday, 20 February 2008 9:01 a.m. To: 'Chris Chamberlain'; dilip.cheerala@nzqa.govt.nz; John Fisher; James Bull; Duncan Tamati; Sandra Monk Cc: David Sharp; Craig Debenham; Dessoulavy Eric; Grant Cathro Subject: RE: Link to demo of Code Value Validation service
Hi
I think this has fantastic potential. Is it possible to be able to read the different standards – it feels odd to be comparing my code set with something that can’t be seen?
Thanks for the work you have done on this – I am very impressed with the way you have researched the possibilities offered by the semantic web ideas and also created something useful from them.
Robyn Smits
Manager, Data Management Unit DDI: (04) 463 2835 Mobile: 027 716 6282
Original Message
From: Chris Chamberlain [mailto:Chris.Chamberlain@tec.govt.nz] Sent: Tuesday, 19 February 2008 11:13 a.m. To: dilip.cheerala@nzqa.govt.nz; John Fisher; James Bull; Duncan Tamati; Sandra Monk Cc: David Sharp; Craig Debenham; Robyn Smits; Dessoulavy Eric; Grant Cathro Subject: Link to demo of Code Value Validation service
I've now set up a public demo of the CVC web service using the Ethnicty 'standards' proposed in my case study, so please give this a try when you get a moment....
http://demo.artisan.co.nz/cvc-demo/up
I've enclosed a couple of business system code value extracts (ex TFS and Basil) for you to test it with.
If you want to try some of your own data just dump it into a CSV file ( comma separated, no quotes ) with column 1 the Code and column 2 the Value (any other cols will be ignored.) upload and let me know how it looks 
A couple of known issues:
- characterset conversions are not handled fully yet - you may see something like "M\xe4ori"
- the percentages shown in the ranking view ( select Standard:All ) are a little strange, but I think the ranking order itself is OK.
Any feedback, ideas for improvement will be most welcome ;).
FYI - things I'm planning already are:
1) A simple dump of each CVC standard in HMTL ( for people ) and in genericode XML for machines.
2) some user control over the degree of 'fuzziness' to apply to the value matching.
3) A UI to support discovery, searching and navigation across the cvc:hasEquivalent and cvc:hasWiderDefinition relationships (to help demonstrate the power of the traceablity/mapping features of CVC).
Kind regards
DILIP 22 Feb
attachment:nzqaethnicitycodevalues.csv
Hi Chirs,
The web service looks so simple, but I can realise how much work that went in - well done.
Attached is the nzqa code value list that we use. I have tried it with the attached file and analysed the results. Please not the following items are not a criticism, I know there is still a lot to be done and this is only an initial trial. But I thought I should share my understanding and comments with you.
Comments/confirmations/queries:
I am assuming the matching logic your web service uses is based on the pair values (i.e. code+value)
"Missing code values" means the listed code values are not in the uploaded file. A better description would be "The following code values do not exist in the < uploaded file name>.
"Extra code values" means additional code values that are in the uploaded file.
"Match" means both code and value from the uploaded file matches with the reference file.
Test rankings:
- Quality refers to match quality of the supplied or referenced code value list? - How can the quality be negative, the least should be zero. (Not sure of the algorithm)
I think the web service should provide more functionality than just straight forward matching. In most cases either the code or value match. For example 31 Tongan (nzqa list) and 331 Tongan (ETHSDR07) the value is same but the code is not. If we were to exchange data based on the nzqa code value list then you have to map 31 to 331 because the it is a valid value and both list mean the same in this instance. The web service should be able to match such items and provide a reasonable mapping. May be you could use the has_narrower or has_wider relationships and resolve this type mapping. I guess you may have these items in your plan to implement in the next phase.
Cheers,
Chris Chamberlain
IT – Data Management
Data Architect