High-Level Flow Model for Anomaly Detection in Object-Oriented Programs

Abstract

In this paper, a High-Level Flow model representing flow information of an object-oriented (OO)program for data flow analysis is presented. This model introduces to conventional data flow additional properties which are made by object abstraction, encapsulation, inheritance, and polymorphism. In an OO problem, an object encapsulates both component data (attributes) and operational functions (methods). The model employs method path expression to represent the flow information of a method, where a method path expression is a set of path expressions of which each describes the accesses of an attribute in the method. The flow information of an OO program can be derived by concatenating method path expressions related. Due to OO features, there are two additional types of data flow anomalies, a message-sequence anomaly and a class-definition anomaly, described in this paper. These anomalies can help indicate programming errors for debugging an OO program. The algorithms of detecting these anomalies can be implemented efficiently with bit-patterm computation.