Abstract
Search personalization aims to tailor search results to each specific userbased on the user's personal interests and preferences (i.e., the userprofile). Recent research approaches to search personalization by modelling thepotential 3-way relationship between the submitted query, the user and thesearch results (i.e., documents). That relationship is then used to personalizethe search results to that user. In this paper, we introduce a novel embeddingmodel based on capsule network, which recently is a breakthrough in deeplearning, to model the 3-way relationships for search personalization. In themodel, each user (submitted query or returned document) is embedded by a vectorin the same vector space. The 3-way relationship is described as a triple of(query, user, document) which is then modeled as a 3-column matrix containingthe three embedding vectors. After that, the 3-column matrix is fed into a deeplearning architecture to re-rank the search results returned by a basis ranker.Experimental results on query logs from a commercial web search engine showthat our model achieves better performances than the basis ranker as well asstrong search personalization baselines.