Suppose that you have clickthrough data in the following form -- (query, clicked url, frequency). I wonder if there is any way of using the data to train a ranking function. Naively, you can treat each click as a ground-truth url and frequency as some kind of weight -- that is, frequently clicked url is treated as a more important training instance than a less frequent one. But, it seems too naive and I wonder if there is any other better method.

I am aware of the famous paper by Thorsten Joachims, 'Optimizing search engines using clickthrough data'. The paper however assumes a different form of clickthrough data -- (query, a list of documents ranked by search system, a set of clicked URLs). So, I cannot directly apply his method to my clickthrough data.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.