- Table of contents
- Data integrity
- Check dates
- Check links
- Check link duplicates
- Check similar names
The quality of data is very important to us. Although ultimately the quality and responsibility of entered data lies with the editors and managers of the project we take great care to avoid entering of inconsistent data on a technical level, e.g. with the user interface it is not possible to enter begin dates which are later than end dates.
Nevertheless mistakes can happen not only on the application level but also e.g. when importing data from other projects or a file which metadata is already in the system is deleted outside the application.
Since data integrity is important for the quality of research data we implemented functions to check possible inconsistencies which are available at the admin menu entry and are described in detail below.
In this view entries like dates which are not linked are shown. They are possible artifacts from imports or bugs and can be deleted. If they seem to appear regularly (without imports or known bugs) please report that issue.
Entities without links¶
The entries shown have no links. Maybe thats OK but maybe they are artifacts or were forgotten to link.
These types where created but have no sub types or associated data. Maybe they originate from the first install or where never used.
Here are listed file entities which have no corresponding file, most likely because this file doesn't exist anymore.
Files that have no corresponding entity are listed here.
A check if an entity is linked to itself. This could happen e.g. if a person is married to herself or a type has itself as super. It shouldn't be possible to create circular dependencies within the application. Nevertheless it's a useful check for e.g. if data is imported from other systems.
In this view invalid date combinations are shown, e.g. begin dates which are later than end dates. These entries should be cleared up because otherwise, they cannot be updated because the user interface won't allow saving entries with invalid date combinations
With this function every link will be checked for CIDOC CRM validity. Depending on the amount of data this could take some time. Date entered with the OpenAtlas user interface should always be CIDOC CRM valid but in case you e.g. imported data you should use this check afterwards.
If invalid links are found they should be dealt with outside the application.
Check link duplicates¶
There are actually two checks:
- The first one checks for duplicate links which are identically and can be safely deleted when clicking the Delete link duplicates button.
- In case the first test found no duplicate links it will be checked for entities connected multiple times two a type which is defined for single use.
E.g. a place has the type castle and city. In this case you would only see one in the user interface and the other one would get deleted in case anybody updates the entry.
Here you have the option to look at these and remove the wrong ones clicking on the Remove link beside the entries in the last column.
Both checks shouldn't find anything wrong with data entered with the application but nevertheless it could happen because of imports or unknown bugs.
Check similar names¶
Here you can search for similar names. Depending on selection and data volume this might take some time.
- Classes: select the class which you want to search for similar names
- Ratio: select how similar the names should be. 100 is the default and means they should be absolute identical.
The lower you set the number the more names which are similar will be found but it will also get more time consuming, so you should begin with a higher numbers.