Hierarchical similarity measurement model of program execution

The concept of similarity measurement is widely used in a number of research areas such as data mining, software testing and debugging, and software security. Several researches have been conducted to continuously improve and explore for a better and new measurement of similarity. The similarity measurement mostly deals with data, items, or features of a program. However, there is no similarity measurement for output of program execution which can be various data types including primitive, abstract and heterogeneous data types such as numeric, character, list, tree, table (relation) and object. Lack of these measurements can reduce the effectiveness of software testing techniques or the quality of vulnerability detection methods to a certain degree. This paper proposes hierarchical similarity measurement model for measuring program execution. The measurement is categorized and arranged into successive levels according to its data type, and its calculation for a higher level data type is performed based on lower level data types. Some operations performed on our similarity measurement are briefly introduced. In addition, we present a similarity graph that can be used in an optimization process.