I've just embarked on a project to recommend similar candidates to employers for a website that offers employers the ability to search its CV database.

My background is in engineering research on biomedical images - from simple predefined feature extraction to CNNs.

I'm about to start with a simple bag of words approach standardised by the frequency of appearance and see where that gets me.

Has anyone done anything similar before and has suggestions for which algorithms might give good results?

The data is very well structured, separated into seperate sections for each previous employment etc.


1 Answer 1


Try content based filtering, you can cluster candidates based on similarity of keywords and by assigning similarity scores. Check tf-idf/cosine similarity measures. Additionally you can customize your algorithm by giving extra weight for section wise similarities since your data is very well structured.

Also check collaborative filtering on Mahout.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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