A new approach based on ants for solving the problem of horizontal fragmentation in relational data warehouses

The subject matter falls within the context of optimization of relational data warehouses. It involves using the algorithm based on ant colonies for the selection of horizontal fragmentation which is one of optimization irredundant techniques. The character NP-complete characterizing the selection of this technique justifies the use of approximate methods or "meta heuristic” to solve it in a finite time. Indeed, the collective intelligence of artificial ants in solving combinatorial optimization problems NP-Complete is a very promising activity. This approach inspires its capacity through the transfer of learning within the colony in a manner which uses the stigmergy for communicate the choice of good solutions based on visibility and the deposit of pheromone. In this article we have modeled our problem of selecting a horizontal fragmentation scheme that be supported by the approach based on ant colonies while defining the input variables which are: the unfragmented data warehouse, the query load frequently used and the maximum number of fragments required by the administrator of the data warehouse (ADW). The result output is the horizontal fragmentation pattern that minimizes the overall cost of the load of requests. The success to formalize the problem as a knapsack problem permits us to present a new approach for resolving the horizontal fragmentation problem. Experimenting with our approach using a Benchmark (APB1 in our case) is one important way to verify the effectiveness of the proposed method on the one hand, and the power to relate to other methods that exist in this area, on the other.