...
Individually mapping variables can be massively time consuming if there is no existing documentation, and if variables do not match question names. Therefore the following rules have been applied to the mapping:
- Each item (question item, variable or question grid) can only have ONE topic mapping. If the item is such that two topics must be applied, the mapper moves up to the top level of topics or declines to give a topic.
- Items related to one another during the question-variable mapping must have the same topic. Again, if this does not work, the mapper moves up to the top level of topics or declines to give a topic.
...
- .
...
- Do the topics available cover all of the items that need to be mapped?
- Is mapping only one topic viable? Are there alternatives and if there are what would the practical application of them take?
- Is enforcing the relationship mapping viable? Are there alternatives and if there are what would the practical application of them take?
A constant challenge of the CLOSER project to enhance historic metadata is to ensure that the solutions it arrives at maintain a balance between being as detailed as possible whilst still being both accurate and practical. These three requirements - detailed, accurate, practical - are in tension with one another. To ensure detail is accurate requires resources. The more detailed this becomes the greater the resource required to ensure accuracy. At no point in this project do we wish to be inaccurate and the project has a finite resource. Therefore it is important to maintain control of the detail and ensure that the desire to be extremely detailed does not risk either accuracy or practicality. Topic mapping is perhaps the area where this tension is most obvious. Additionally, the work done needs to be usable - if it is not detailed enough, it will have no use. If it is too detailed (i.e. the number of topics was too large) it would become unwieldy from a user point of view.
...