Program Readability: A Proposed Software Metric
Paul S. Fisher, Jinsuk Baek, Minho Jo · 2009
This paper addresses the issue of readability extended from the ideas of the Flesch Reading Ease only it proposes another approach that is based upon pattern recognition, and a measure of the locality of these patterns as exhibited in the program text. We suggest that short patterns are easier to grasp than patterns that are longer. We test this technique using a pattern recognition system we have developed called Finite Inductive Sequences, and apply it to English text to determine if it matches the Flesch Reading Ease score which intuitively it should match. Then we apply it to C++ programs, one written by an expert programmer and the other written by a competent programmer, but not an expert. We then derive comparative measures for these two programs, suggest some ideas related to these preliminary results, and suggest some additional considerations for this kind of work. The application of this would primarily be associated with development in the Agile environment, but it could be appropriate for any developmental or maintenance activity involving programming.