A hybrid click model for web search
Danial Bidekani Bakhtiarvand, Saeed Farzi · 2019
Annually, web search engine providers spend more and more money on documents ranking in search engines result pages (SERP). Click models provide advantegeous information for ranking documents in SERPs through modeling interactions among users and search engines. Here, a hybrid click model is introduced by combining a PGM-based and a neural network click model. Hybrid click model tries to predict users' clicks behavior on the documets which are represented in SERPs. Indeed, a weighted k-nearest neighbors has been employed to provide final decision based on UBM and LSTM click models scores. The proposed system is evaluated on the Yandex dataset as a standard click log data set. The results demonstrate the superiority of our model over the state-of-the-art click models in terms of perplexity.