Thinking of self-submission of one’s own work into a repository and the quality of the resulting metadata, particularly subject metadata …

Any cataloguer who has ever tried to catalogue a technical report which helpfully includes user-submitted keywords will be conversant with the potential inadequacies of such keywords. It is not unusual, in my experience, for reports on highly complex subjects to include only vague keywords, such as “aspects”, “experiments” and “research”.

It is possible to build up highly complex controlled vocabulary infrastructures into repository software, so different user groups can access their own specialised vocabularies; which is great as far as that goes. But sometimes freeform keywords are the best option in a particular case. So how best to encourage users to choose high-quality, relevant keywords, without having to provide them with pages-long advice?