The topic is turning clickstream data into relevance labels to train machine learning models to optimize search relevance. It’s a fun topic, as it’s far more complicated than just “label things clicked more as more relevant”. There’s dozens of reasons how/why users click things (or don’t click things) besides relevance you have to overcome to derive reasonable relevance labels.
The topic is turning clickstream data into relevance labels to train machine learning models to optimize search relevance. It’s a fun topic, as it’s far more complicated than just “label things clicked more as more relevant”. There’s dozens of reasons how/why users click things (or don’t click things) besides relevance you have to overcome to derive reasonable relevance labels.
1 - http://aipoweredsearch.com