An information coding-based data complexity model

Data complexity analysis should play an important role in software engineering. Unfortunately, it has been relatively ignored in the past. In this paper, we develop an innovative theoretic model called `information coding-based data complexity´ (ICDC) for measuring data complexity. We first define the concept of `information-describing program data´ and derive general formulas for information in various data structures. In order to avoid conflict with software practice, we then advocate two basic laws of data processing. These laws refer respectively to the information-correlation and repetition-removing principles which, we believe, reflect the human information processing mechanism. The complexity of data is defined as the measurement of coded information by eliminating the correlated and repetitive parts. A formal description of data is advanced, from which the correlated information can be calculated. Properties of data based on the ICDC model are also presented, which coincide well with empirical software knowledge