Web search engines gather information from the queries performed by the user in the form of query logs. These logs are extremely useful for research, marketing, or profiling, but at the same time they are a great threat to the user's privacy. We provide a novel approach to anonymize query logs so they ensure user k-anonymity, by extending a common method used in statistical disclosure control: microaggregation. Furthermore, our microaggregation approach takes into account the semantics of the queries by relying on the Open Directory Project. We have tested our proposal with real data from AOL query logs.