Abstract: The SPRINT was improved by introducing a dynamic data structure so as to resolve the problems that SPRINT makes use of multiple attribute lists to store data which occupies system resources, splits were carried out by way of establishing hash table, and the node splitting process is relatively complex. The example result shows that both the spatial complexity in attribute lists and the temporal complexity in the manipulation of split node have been decreased by the improved SPRINT. Finally, SPRINT’s working process was discussed under distributed environment.

Key words: decision tree, SPRINT algorithm, distributed environment