Over the past two years we have been working with CETIS and the HEDIIP Programme to develop a new subject classification system for Higher Education courses to replace the existing JACS system. Classification of courses and course subjects has been one of our main streams of work since we started the company, first with the Learndirect Classification of Subjects (LDCS), then JACS and the Sector Subject Area system.
Although it may not be the most exciting topic, accurate classification of courses’ subjects of study has some surprising uses beyond just data crunching and funding. As part of one of the XCRI-CAP projects looking at how the many and varied potential uses for XCRI-CAP, we developed (with our friends at Intelligent Global Solutions Ltd) a demonstrator for a concept-based course search tool, which was developed into our Course Explorer service.
Most readers will probably know that the majority of search engines (particularly site specific search engines) are based on a string search – that is, you put in a string of characters, for example “psychology” and the engine goes looking for that string and presents back to you all the pages that contain that string. Concept-based searching is quite different – you still enter your string, but the search engine understands that “psychology” is a concept related to a number of other terms and brings back all the pages that contain something related to that concept. So it might bring back pages on Freud or Cognitive Behavioural Therapy as well as anything that mentions psychology specifically.
There is growing interest and capability in concept-based searching for these reasons. The major search engines use concepts as just one of their tools to try and get us the web pages we’re looking for, but even some other course searches are starting to use concept-based searching to pull out more accurate results for students. Our Course Explorer search is actually based on an extensive thesaurus of subject terms from a variety of classification systems we have worked with over the years, and has recently been updated with the new Higher Education Classification of Subjects (HECoS) terms from the project I mentioned above, but others may be based on other systems.
How to tell if your course search is concept-based or string-based
For this test you need a dedicated course search functionality (not a full web search) and a term which you know you have some courses in, but which you also suspect might be using course descriptions for other reasons. “English” is generally a good term to use: most HE providers have some kind of English course, but also have English requirements for international students or requirements to submit applications in English specified on the course page.
The test itself is simple: go to your course search, put in your term and see what comes out. I’ll be using the OU as an example because they use our concept search, but also have a string search on their site (filterable for just courses) to use as a comparison.
Typical string results
courses listed that have nothing to do with your term
courses in no particular order, for example relevant and irrelevant courses jumbled up together in the search results.
The first few results from OU’s string search for English look like this:
There is an extraneous page which isn’t a course page because this isn’t actually their course finder, but only the fourth entry on the list is a course on Science: human genetics and health issues. I’m sure English is involved in working on this course, but this is not a course about studying English.
Typical concept results
courses should be relevant to the term
courses may be in order of relevance with courses which use the exact string first.
The results from the OU concept search look like this:
I’ve only included the UG results to compare with the UG prospectus filter on the string search, but these are clearly more relevant. No science courses, and it’s picked out “Language Studies” and “Arts and Humanities” which do involve studying English, but don’t mention the term. This search is weighted towards English language rather than literature, but if you put in the term “English literature” you get:
which is weighted with literature courses first.
So now you should have a good idea of whether your course search is string- or concept-based and what kind of accuracy your prospective students can expect when they go looking for courses on your site. String searches are improving all the time and yours may be perfectly fit for purpose, but if you would like to learn more about our concept-based search tool take a look at http://www.alanpaull.co.uk/#!course-explorer/phrei or contact Alan on email@example.com.