The major objective of this project was to help user for taking up the next topic in a course. It is needed only in case of multiple explanation or teaching methodologies and learning styles are involved; which is very generalised so to speak.
The concept diagram is as below:
We solved the problem with cosine similarity for finding nearest neighbours for queried user. The nearest neighbour is in the sense of vectorised attributes which are selected and generated by us in random yet logical sense. We have topic ids and user ids for recommendations.
CURRENT STATUS:
We are trying to integrate sli_rec model as validation module which gives accuracy of a particular recommendation.
As can be seen below the 6 paths of the same user.
The first one is the original path and next 5 are also the same but a node is added to the bottom for the recommendation which is inferred from 5 nearest users depicted by red node and edge.
