Abstract: Based on the uncertain data representation model, we proposed an uncertain data generation algorithm ACUDGen (attribute level continuous uncertain data set generation algorithm) for attribute level uncertain data. By introducing the outlier detection algorithmLOF (local outlier factor) algorithm, the algorithm used the outlier factor of each data object as the parameter to control the perturbation range of uncertain data objects, which could well satisfy the distribution characteristics of the original data and solve the problem of lack of the distribution characteristics of the original data in the present work. The experimental results show that the uncertain data set generated by the proposed algorithm has a better clustering effect, and reduces the influence of outier on the clustering results, so that the size of each data object MBR (minimum bounding rectangle) can be adaptively changed according to its own distribution characteristics.

Key words: outlier factor; ACUDGen algorithm, uncertain data, representation model