Research On Algorithms For Detecting Outlier Spatial Lines Based On Topological Relations For GML Data

Outlier detection is an important research problem of data mining. The purpose of outlier detection is to discovery the unexpected, interesting and useful patterns of further analysis. Spatial outlier detection is aimed at detecting the spatial objects which are different form other spatial objects in their spatial attributes and topological relationships. Now, only point objects without line objects or polygon objects are considered in the existing spatial outlier detection algorithms that the different degree on topological relationships are not included in either. Therefore, in this paper, algorithms for detecting outlier spatial lines based on topological relationships for GML data are deeply studied, and some innovative contributions are achieved as follows:1. Algorithms DOL_IR1 and DOL_IR2 are presented for detecting outlier lines based on intersection relationship for GML data. Intersection relations between spatial lines and other spatial objects are computed. The difference degree between one line and another line is defined, In algorithm DOL_IR1, the difference degree is looked as the standard of the distance between one line and another line, Algorithm DBSCAN is used to detect outlier lines based on intersection relationship. In algorithm DOL_IR2, the spatial lines are clustered according to their different degrees and whether the cluster is 'outlier' or 'normal' is decided by its outlier factor. The experimental results show that algorithm DOL_IR1 and algorithm DOL_IR2 both can detect outlier lines based on intersection relationship accurately and effectively.2. Propose algorithms DOL_AR1 and DOL_AR2 for detecting outlier lines based on adjacent relationship for GML data. The difference degree between one line and another line on adjacent relationship is defined. The difference degree is used as the standard of the distance between one line and another line, algorithm DBSCAN is used in algorithm DOL_AR1 to detect outlier spatial lines. In algorithm DOL_AR2, cluster spatial lines by the difference degree on adjacent relationship and define the outlier factor of every cluster, the outlier factor of the cluster determines whether the cluster is 'outlier' or not. The experimental results show that algorithm DOL_AR1 and algorithm DOL_AR2 both can detect outlier lines based on adjacent relationship accurately and effectively. But the runtimes of the algorithms are computed and compared in the content, algorithm DOL_AR2 is more effectively.3. Present algorithms DOL_IA_R1 and DOL_IA_R2 for detecting outlier lines based on intersection and adjacent relationships for GML data. The intersection relationship and the adjacent relationship constitute the topological relationship. Algorithms DOL_IA_R1 and DOL_IA_R2 consider the topological relationship and define the different degree based on intersection and adjacent relationships. Algorithms DOL_IA_R1 and DOL_IA_R2 detect outlier lines using algorithm DBSCAN and clustering algorithm. The experimental results show that algorithm DOL_IA_R1 and algorithm DOL_IA_R2 both can detect outlier lines based on intersection and adjacent relationships accurately and effectively, the efficiency of algorithm DoL_IA_R2 is higher.