DEFINITION Personalizing or recommending OLAP queries aims at making the OLAP user experience less disorientating when navigating huge amounts of multidimensional data (also called cubes). Such approaches allow coping with too many or too few query results, or suggesting new queries to pursue the navigation. Personalization allows adding preferences to a query for filtering out irrelevant results or ranking the results to focus on the most relevant first. It also allows turning selection predicates (hard constraints) into preferences (soft constraints) to favor non-empty answers. On the other end, recommendation allows to leverage the cube instance and/or past navigations on it to complement the current query result. The general problem can be formally defined by: given a sequence of queries S= (a session from now on) over an instance I of a cube schema C, a user profile P (consisting of ordered multidimensional objects), a set of past sessions L (a log from now on), generate a set of one or more queries Q={q p 1 , … q p n } such that, typically:  The queries in Q are sub-queries of q c (personalization), in the classical sense of query inclusion, or none of the queries of Q are sub-queries of the queries of S (recommendation),  The queries in Q maximize an interestingness score, In this definition, S represents the current session, with q c the last query of this session (the current query). HISTORICAL BACKGROUND OLAP Personalization and recommendation approaches are distant descendants of cooperative database [11] techniques aiming at enhancing database management systems with a cooperative behavior. Cooperation can be introduced at the different stages of the retrieval process, which is typically iterative. The purpose of the cooperation includes: helping the user to formulate a query corresponding to an objective and acceptable by the database system, dealing with empty answers or too few results, or suggesting additional information and explaining the query result. This retrieval process perfectly reflects the activity of OLAP users, who interactively analyze multidimensional data, often without exactly knowing what they are looking for. OLAP queries are normally formulated in the form of sequences called OLAP sessions, by using basic operations to transform one OLAP query into another, so that the new query gives a better understanding of the information retrieved so far. The huge number of possible aggregations and selections that can be operated on data may make the user experience disorientating, and OLAP