The document discusses dynamic user profiling for search personalization. It begins by describing classical search systems that return the same results for a given query regardless of individual user preferences. The research aims to enrich user profiles through dynamic group formation and temporal awareness to improve search personalization. For dynamic group formation, it constructs query-dependent user groups using latent Dirichlet allocation to model topic distributions. It then enriches individual user profiles by averaging over group profiles. For temporal profiles, it builds profiles that decay older document relevance over time to better reflect current interests. Evaluation on search logs shows the approaches outperform baselines by improving metrics like mean average precision.