dbpedia
Field Sampling in mapping server
Enhancement to the mapping server:
- On the template stats page (eg this)
- for every field, there is a blue number of occurrences
- make that a hyperlink to show some of the occurrences (say 100)
Why:
- Extremely useful to understand the meaning of some fields, since many fields are not well documented in the wiki templates
- And to see whether they're links, text, or both (re Object/DataProp Dichotomy)
What:
- Make two columns: page, field.
- Would be great if you output HTML, but TDV would also work fine.
- First col is a link to the page in Edit mode
- Eg if I click on "venue":
| page | field "venue" |
|---|---|
| Alpine skiing at the 2002 Winter Olympics | `[[Snowbasin]] (downhill, super-G, combined), [[Park City Mountain Resort |
@VladimirAlexiev I am working on this task. Added a new page where we will list the field occurences, link to my code
To proceed further, I need a guidance on how to get all occurences of field.
Hi @ujjwalwahi! I don't know where to get them from, I'm sure @jimkont can help you. Where is the code of server/templatestatistics that shows the number of occurrences of each field?
I did a bit of tracing. counter is derived from count which is obtained from sortedProps which comes from getMappingStats() which gets them from mappedStatistics. Search shows it in MappingStatsHolder. The closest to what we need is propertyUseCount Again search shows where propertyUseCount is summed. Gets it from val properties that's a map (name, (count, mapped)) Search shows that's constructed here and is obtained from wikiStats.templates This seems to be loaded here from some file...
This is as far as I got. @jimkont or @jcsahnwaldt can help better.
@VladimirAlexiev I think, they're loaded from here. Yes, that's the same place as ignorelists. I guess you've already been there recently :) There's even a hint just one line above the last one you linked to.
So @ujjwalwahi this is a bit of a dead end: the extractor loads field mapping stats from files, but I still don't know HOW these files are produced. Maybe @jimkont or @jcsahnwaldt can help?