It has become increasingly clear to me that it is important to position the skills construct in relation to psychological constructs.
Because skills are defined primarily by their result, and situations are (often) highly diverse and unpredictable, skills are inherently ideographic/ qualitative/ empirical. However the people performing the work find ways to recognize situations and predict outcomes, and develop their own paradigmatic constructs using language.
The psychology profession has spent the last 100 years systematically building nomothetic, quantitative variables to describe human behaviour in ways which are predictive, but also in ways which rely heavily on language; the personality systems like Hexaco, Big 5 etc. were all developed lexicographically. Their big limitation is that you can never really know 'what people mean when they say words'.
To convert skills into skills data, you have to know what workers mean when they say words. You have to observe the work, understand how that is being converted into language, then apply systematic approaches to coding that information into data systems. A lot of that information concerns human behavior, which can be coded using established psychological constructs...which then allow for predictive work to be performed.
However you can also learn the paradigms workers have developed to recognise situations and types of work, and code those systematically into systems of situations.