Comparing and integrating of different data structures e.g. relational databases of information systems is a current problem in information sciences. Various solutions have appeared in the last 10 years aimed to achieve a high accuracy level in schema integration and similarity measurement of entities originating from different schemas. The capabilities of approaches are usually properly evaluated from the point of view of accuracy. However the computational complexity of the proposed algorithms is hardly ever examined in these works. We claim that efficiency of a proposal can only be measured by taking into account both the accuracy and the computational requirements of participating algorithms. Since there are many known measurement methods and metrics for the evaluation of accuracy, the focus is set for the analysis of their computational complexity in this paper. After the problem formulation the main ideas behind our method are presented briefly. Many kinds of approximation techniques and applied algorithm theory are used to evaluate different approaches. Three specific approaches were also selected to present the work of our method in details on them. Experiments run on various test inputs are also included.
展开▼
