Benchmarking XML-Schema Matching Algorithms for Improving Automated Tuning

Several matching algorithms were recently developed in order to automate or semi-automate the process of correspondences discovery between XML schemas. These algorithms use a wide range of approaches and matching techniques covering linguistic similarity, structural similarity, constraints, etc. The final matching combines arithmetically different results stemmed from these techniques. The aggregation of the results uses often many parameters and weights to be adjusted manually. Generally, this task is achieved by human experts and requires a perfect understanding of the matching algorithm. In order to reduce the human intervention and improve matching quality, we suggest automating the tuning of the various structural parameters used within XML-Schema matching algorithms. In this work, we offer a benchmark, for three tools, that seeks mathematical relations between parameters values and schema topology. In consequent, we propose an algorithm for the tuning of these parameters for studied tools.