CiDHouse: Contextual SemantIc Data WareHouses Selma Khouri 1 , 3 , Lama El Saraj 2 , 4 , Ladjel Bellatreche 3 , Bernard Espinasse 2 , Nabila Berkani 1 , Sophie Rodier 4 1 ESI, Algiers, Algeria (s khouri, n berkani)@esi.dz 2 LSIS, Marseille, France (firstname.lastname)@lsis.org 3 LIAS/ISAE-ENSMA, Futuroscope, France (selma.khouri, bellatreche)@ensma.fr 4 Assistance publique des Hpitaux de Marseille, France (lama.elsaraj,sophie.rodier)@ap-hm.fr Abstract. Dealing with contextualized data is a key challenge in different fields of infor- mation systems, databases and data warehouses (DW). Nowadays, DW Systems are often mono-context. However, in real life applications, DW indicators are shared by many users from different profiles. For example, in the medical domain, users can be doctors, researches, nurses, computer scientists, etc. They need comprehensive results adequate to their context. To tackle this challenge we propose to use ontologies to treat contextualization problem at the semantic level. Thereby, we offer an approach that makes DW systems multi-contextual and able to personalize results based on user’s context. In this paper, we present a novel approach based on ontologies for designing multi-contextual DW. We propose an ontology formalism that incorporates the contextualization concepts. We specify our approach that takes contextualization into account from the conceptual level. We validate our proposal using a real case study from the medical domain. 1 Introduction Nowadays, the data warehouse technology becomes an incontestable technology and tool for busi- nesses and organizations to make strategic decisions. Data warehouse (DW) is considered as a pillar of the integration industry widely developed in last two decades and recently in the big data field [7]. A DW is defined as a stepwise information flow from information sources owned by het- erogeneous services and departments through materialized views towards analyst clients. One of the difficulties of building a DW application is handling the heterogeneity of information sources. Ontologies play an important role to reduce this heterogeneity and to reason on the ontological concepts. Note that domain ontologies have been widely developed in several domains such as medicine, environment, engineering, etc. This development motivates the DW community to con- sider it in the design steps of the warehouse applications, especially in the conceptual modeling and ETL phases [22,4]. During the conceptual phase the ontologies may represent the global schema of the DW [3, 6]. Some other works proposed then to attach an ontology (usually called local ontology) to each source participating in the DW contruction, and to define mappings between the global and local ontologies [14, 25]. The DW considering sources embedding ontologies in their repository (usually called semantic databases (SDB)) correspond to this architecture. Two integra- tion scenarios can be defined: (i) correspondences between global and local ontologies are defined a priori at the design time of SDB. In such case, the integration process is simply assimilated to an integration of mappings. Designers agree to make efforts when designing the sources in order to get a ‘free ’ ETL process. (ii) Correspondences are discovered a posteriori either manually or automatically, which is an issue related to the domain of schema and ontology matching/alignment. Once the mappings discovered, the integration process resembles to the first scenario.