Article Preview
TopIntroduction
Data warehouses (Chaudhuri et al., 2011) are more and more becoming a key component of data-intensive systems, like, for instance, recent Cloud environments (Buyya et al., 2011), where data explosion plays a critical role and whose management represents a major research challenge (Hsu et al., 2011). Due to this evident relevance, knowledge discovery and management (e.g., (Cuzzocrea, 2009)) represent a viable solution to the problem of enhancing the way we actually access, manage and explore large amounts of multidimensional data stored in data warehouses. Supporting advanced query answering and actionable knowledge extraction from data warehouses can thus be reasonably intended as one of the most challenging issues for next-generation data warehouse research. This problem is also directly related to OLAP since it represents the most popular interface to data warehouses, which heavily makes use of fortunate metaphors such as dimension, hierarchy, measure, multi-level analysis, multi-granular analysis, and so forth (Chaudhuri et al., 2011).
Data size and management via querying, browsing and exploration primitives in data warehouses are a relevant limitation to the goal of enhancing the user-centric interaction with existing data warehouses. In fact, users are easily disoriented by the size of the multidimensional query space induced by data warehouses, which is usually very large (e.g., (Cuzzocrea & Serafino, 2009)), and by the size and the semantics of retrieved results (e.g., answers to OLAP queries). This also poses severe limitations to the problem of efficiently performing analytics over multidimensional data stored in data warehouses (Cuzzocrea et al., 2011).
These problems have been recently investigated via two major approaches. The first one proposes the so-called data warehouse personalization (e.g., (Golfarelli et al., 2011)), whereas the second one pursues the alternative idea referring to the so-called data warehouse recommendation (Giacometti et al., 2011; Bentayeb & Favre, 2009). Data warehouse personalization suggests to support user-centric interactions by taking into account user’ preferences and needs (Stefanidis et al., 2011), also focusing on user groups. Data warehouse (DW) recommendation instead aims at supporting user-centric interactions by directly assisting users by means of suggesting new contents, for instance in terms of query results or refinements, and providing decision making primitives (e.g., (Giacometti et al., 2009)).
As orthogonal to these axes, some paradigms seem to become relevant during next years for both data warehouse personalization and recommendation. Among them, we recall the need for supporting novel kinds of query paradigms against data warehouses, such as cooperative queries (e.g., (Cuzzocrea, 2008)), collaborative queries (e.g., (Khoussainova et al., 2009)), intensional queries (Motro, 1994), keyword-search queries (e.g., Blunschi et al., 2011), and so forth. At the same time, supporting on-demand actionable knowledge on data warehouses that looks for knowledge patterns instead of data, plays a “first-class” role.