| Popis: |
This paper presents an unsupervised, content-based approach to match users with shorter pieces of specific learning content, lectures, to target their learning goals at a more granular level. This method is especially useful when implicit data is unreliable or limited. At a high level, our approach generates a set of lectures for every topic via clustering and then matches lectures to users via users' topic affinities. Our central hypothesis is that important, fundamental concepts are repeated within many courses on the same topic, and by extracting clusters, we can identify these key information lectures per topic. |
