Closed csarven closed 9 years ago
Check whether it is possible (i.e., reliably) to build the metadata (e.g., DSDs, concepts, codes, properties) programmatically, instead of hand coding
Okey, I'll try that
2014-12-09 10:15 GMT+01:00 Sarven Capadisli notifications@github.com:
Check whether it is possible (i.e., reliably) to build the metadata (e.g., DSDs, concepts, codes, properties) programmatically, instead of hand coding. Do it in several steps if necessary e.g.:
- Tooling extracts as many terms as it can
- Manually enhance the descriptions
— Reply to this email directly or view it on GitHub https://github.com/csarven/doingbusiness-linked-data/issues/1.
I wrote a script to generate the DSD file. Should I push it into the same directory (scripts/) where the actual scripts for the ETL will be?
Yes.
Has been pushed to https://github.com/csarven/doingbusiness-linked-data/tree/dev/scripts
2014-12-15 16:19 GMT+01:00 Sarven Capadisli notifications@github.com:
Yes.
— Reply to this email directly or view it on GitHub https://github.com/csarven/doingbusiness-linked-data/issues/1#issuecomment-67008925 .
re: https://github.com/csarven/doingbusiness-linked-data/commit/27c94e0a093c0c569f28f58813b762deb2af7228
Oh my goodness!
What I meant was that the script actually extracts the terms for you, so that you don't have to manually list them in the code. Typically, you would tell the script (whatever tool you are using to e.g., extract, map) to look for a particular location in the file, and then generate rest of the stuff.
What you have is a semi-automatic approach, which is an improvement from manually writing each part, but this is still a lot of work and repetition which you should try to avoid.
The point of aiming for automatic generation is so that you don't have to create two completely separate processes for building the metadata (e.g., DSD), and data (e.g., datasets). It is possible that what you have in https://github.com/csarven/doingbusiness-linked-data/commit/27c94e0a093c0c569f28f58813b762deb2af7228 is the best scenario but I would seriously encourage you to look into this again.
For instance, if you use tarql to build the dataset observations, why wouldn't it be possible to build its DSD and rest of the metadata? What is missing or insufficient? Try to address that because it will make your life much easier in the end.
Has been changed.
Check whether it is possible (i.e., reliably) to build the metadata (e.g., DSDs, concepts, codes, properties) programmatically, instead of hand coding. Do it in several steps if necessary e.g.: