User:Marjan Tomki SI/Constraints subpage

Introduction

Data modelling and troubleshooting problems

Constraints are tools to help having logically correct data entered, to verify correctness of data model, and to verify the paths for data retrieval (use of data).

The problem is that extremely often (not only here) data gets entered before the data model is sufficiently good. I did my best to evade that whenever I encountered that problem, but where I could not that had costly (in time, usefulness and at the end also money). Some of the consequences ː

• you can't enter correct, but rare data (new or exotic person names etc.)

- solution ː data entered provisionally, to be confirmed by an expert and appropriate procedure (e.g. name checked with both the person named and initial registration office, when needed). If it is official and legal in one country, it might be usual (even legally mandatory) to "correct" in another (mis-used historically often to suppress names of some nationalities, but also seen as practical - we can't read that; spell it in our language - see also for documents for foreigners in China). - Let constraint require expert opinion for each occurrence of insertion (otherwise new zillion of misspellings enter data basɛ). I dealt with that successfully.

• you don't have end user (general editor) targeted help, yet, for when a constraint is violated and data can't get entered. It seems it gets better, but I was guessing more than once, and have probably guessed wrong at least once (and possibly more times). At least one of those I corrected (hopefully well) later.
• data gets entered before the model is normalised

- problems ː not only increased burden for data entry (and control and correction), but also inconsistencies with the data (and hard to detect, even harder to clean out). - solution ː normalise and debug the model (with test problem data) as soon as possible, so that at least you correct the model with all the data only once (the problem of population register of Slovenia in roughly 1960-1990, when in the beginning theory of normalisation was not yet known well enough. I got related problems systematically detected, but couldn't prevent them then. I was not in position to initiate development of a new version of SW based on new data model, and there were no general data base SW packages available there and then, yet).

Annoying things here

Readability

This is not as readable as it could, and should, be, as it seems a lot of elsewhere WP formatting works differently here.

One of annoying things here at the moment is that I can't enter the symbol for addition (star * ) because the keyboard combination that enters it in my notepad here enters something (or several somethings) else, because it seems to be used as shortcut for something local here. That goes for both keyboard combinations (on letters and numeric parts of the keyboard). The star seems to be hijacked to enter accent symbols over letters. I can enter what I want by clicking on format bulleted list, but that's a problem when no pointing device is in system.

It also seems to enter another type of colon ː which is not recognised by my spellcheck SW.

Entering links to objects etc here

Cant enter https://www.wikidata.org/wiki/Q533534 for item Cisleithania (Q533534) as internal WD link (neither for page with the problem, nor for value in a statement...). Need to know more to define the problem. Some of links got internal - it maybe goes with Q and P no-s without verbal descriptions, and possibly not with talk pages (or not as I expect at least).

Problem examples

Cisleitania and related things

Comment

As a non-historian, I was looking for citizenship of my grandfather and his brother, who were born in Bukovina (the place is now in Ukraine), got recruited in the army, did their stint and as a honourably discharged soldiers got the opportunity to work at the railway. There personal seem to having been moved where needed, and grandfather got to Slovenia, got married and here I am.

I'll recheck the documents, but I don't recall Cisleithania mentioned there, and I don't see anyone but professional historians now knowing what Cisleithania is (or was), but they understand Austrian side as opposed to Hungarian side. Somehow that should be made better generally understandable.

Problem

For [(Q533534)] there is the statement for property Country (P17) and Austria-Hungary is the value with no flags. Nearby there is another with https://www.wikidata.org/wiki/Property:P131 located in the administrative territorial entity; Austria-Hungary is entered, but flagged for constraint violation ː Austria-Hungary should have a statement country.

• Austria-Hungary has the statement instance of country, but not statement country (because that statement seem to describe which country it belongs to).

Can't yet report properly (I seem not to enter links in expected way). Waiting to when I learn.

Events location

Comment

In Property_talk:P131#Events there was a discussion about locating events in administrative unit P131, and location (P276) was suggested as usually more appropriate (OK by me), but next answer was to move all (P131) claims to (P276) claims. With that I disagree, because several event types are spread over a larger area. Some of those are races (cycling, rally car and motor races, sailing) and probably shouldn't be handled automatically so.