Core schema mappings

Scalable data exchange with functional dependencies

“…However, by using the script generation algorithms described in [17,24], the best we can achieve is to generate a pre-solution, i.e., a solution for the tgds only, as shown in Figure 1.c. It is easy to see how the pre-solution is unsatisfactory from several points of view.…”

“…Contributions The main technical problem addressed in this paper is the following: given a mapping scenario containing a set of source-to-target tgds and a set of target egds, our goal is to generate an executable SQL script that can be run to generate good solutions for the given scenario. To do this: (i) we introduce a best-effort rewriting algorithm that takes as input a scenario with s-t tgds and egds and, whenever this is possible, rewrites it into a new scenario without egds that can be efficiently implemented using an SQL script; in the paper, we show that our algorithm succeeds in many practical cases, including the example above; a key intuition behind the algorithm is that source constraints can be of high value in order to generate solutions that satisfy the required target constraints; (ii) the rewriting takes advantage of a number of novel techniques; among these, a notion of overlap tgds, based on the idea of chasing egds at the formula level to avoid the introduction of unneeded null values, and a sophisticated skolemization strategy; in this way, we significantly push forward the expressibility of our SQL scripts; (iii) then, we investigate the issue of generating optimal solutions, i.e., core universal solutions; we show that the rewriting algorithm to handle egds is modular in nature, since it can be coupled with the core-computation algorithms developed in [17,24]; this process is definitely non trivial, due to the complex rewriting that we use for egds; in this way, we provide a much needed extension to the corecomputation techniques in [17,24]; (iv) finally, the techniques developed in the paper have been implemented in the +Spicy mapping system [18]; using the system, we provide a comprehensive evaluation of the algorithms presented in the paper, to show that they scale very well to large databases, and that they actually generates solutions that are much more compact than those generated by current mapping algorithms. Applications This is the first algorithm that enables the generation of solutions for mapping scenarios with egds in a scalable way.…”

“…We say that a set of FO-rules is a core implementation for M if it is a complete implementation that always returns core solutions for M. We show that the core-computation algorithms developed for scenarios without egds [17,24] can be used as building-blocks to turn a complete implementation Σ …”

“…Articles from this volume were presented at The After the seminal Clio papers [20,21] introduced the key algorithmic techniques needed to generate mappings from correspondences, and executable scripts from mappings, a solid theoretical foundation for the data-exchange problem was laid in the framework of data exchange research [10,12]. More recently, sophisticated algorithms have been proposed [17,24] to take advantage of the theoretical background of data exchange and adopt a more principled approach to the generation of solutions. In fact, a data-exchange problem may have many different solutions.…”

“…Among these, the notion of core solution [12] was identified as the "optimal one", since it is the smallest among universal solutions. Second-generation mapping systems [17,24] made a significant step forward by introducing algorithms that materialize core solutions by using runtime SQL scripts.…”