Comparison of Subjective Entropy and User Estimates of Software Complexity
We investigated subjective entropy, a new information-theoretic measure of program comprehensibility which accounts for semantics and pragmatics involved in programmer-program interaction. Student subjects were administered subjective entropy tests on program samples in dBase III and Lotus 1-2-3 Macro Languages. Since the test employed an automated tool and required an average of only 20 minutes, we found it practical to administer. Subjective entropy scores indicated that the program in Lotus 1-2-3 macro language was more difficult to understand. The scores were consistent with expert opinion and with the subjective ratings of the subjects.
KeywordsProgram Sample Automate Tool Software Complexity Syntactic Property Bell System Technical Journal
Unable to display preview. Download preview PDF.
- Duffy, T. and Duffy, W., 1987, A Casebook: Four Software Tools, Wadsworth Pub. Co., Belmont, CA.Google Scholar
- Zunde, P., 1985, An Information-Theoretical Metric for Testing Program Comprehension, AIRMICS Tech. Rep. RARI-85–1, Atlanta, Georgia.Google Scholar
- Xenakis, JJ., 1987, 1–2–3: A Dangerous Language, Computerworld,June, pp. 31–34.Google Scholar