Daniil Korbut
Aug 25, 2017 · 1 min read

Thanks for the question Jonah!

As I mentioned in my article, the best approach to get started are unsupervised methods. If you have only descriptions of people, you can apply some clustering algorithms to make personal recommendations from cluster. Than, after collecting some feedback, it could be better to apply, for example, CF algorithms in each cluster. That allows to build better recommendations for similar persons.

)

    Daniil Korbut

    Written by

    Deep Learning Researcher at Insilico Medicine