Ranking Datasets by Level of Difficulty

Strictly speaking, datasets for testing software cannot be ordered by level of difficulty. The characteristics that make a dataset difficult for one piece of software will make it easy for another.

However, to help non-experts use the StRD service and correctly interpret the results of tests they run, an approximate ordering of the datasets seemed necessary.

Having datasets rated at a lower level of difficulty helps users identify the tasks that all software should handle relatively easily.

Having datasets rated at a higher level of difficulty offers devlopers a target for making more capable and robust software.