If you look at publications, you can have a dataset

  • title of publication
  • list of authors
  • number of pages
  • year of publication

The Level of measurement of "number of pages" is interval scale, the year of publication is interval scale as well, the title is nominal. But what is the list of authors? Simply saying that it is nominal seems not to capture a major part of this attribute.

  • $\begingroup$ As far as I understand this concept of level of measurement, it's a typology for basic or atomic types of information. This would be the equivalent of primitive types in programming languages, as opposed to complex types such as objects (in OOP) made of multiple sub-components. The typology doesn't seem to account for complex types of data, therefore it doesn't have a term for lists (or any other kind of data structure, for that matter). $\endgroup$ – Erwan Jul 7 at 11:56

With a list such as this list of authors, the only representation which truly capture the semantics is a (long) list of boolean features, one for each possible author in the set. Ideally one would also pre-process the whole set of authors in order to match the different ways a name can be written, e.g. "J. Doe" = "John Doe" = "Doe, John".

  • $\begingroup$ "the only representation which truly capture the semantics is a (long) list of boolean features" certainly not, as a list of lists does also truely capture the semantics. And it is way more efficient. $\endgroup$ – Martin Thoma Jul 6 at 20:37
  • $\begingroup$ Sorry I read the question too fast apparently, I thought the goal was to represent this information as features in a dataset. I read it again and I realized that I don't understand it actually, sorry! I As a side note, considering the title (or any multi-word text) as nominal is a quite big simplification imho. $\endgroup$ – Erwan Jul 6 at 22:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.