kh (09/11/82)
In response to my request for information on research into the occurence and correction of errors, I received these responses. Thanks to all who answered, and if anybody has any more information, I'd be happy to see it. Kenny Hirsch unc!kh ------- There have been a number of studies on the topic; check the indices for CACM for about the last 10 years or so. PL/C in particular does some fairly good error correction; I seem to recall reading of a spelling-error corrector as well. -------- See the article in IJCAI 81 by D. Norman. -------- A good starting point is RIpley and Druseikis collection of Pascal errors in Computer Languages, v 3, 1978 -------- There has been a little research done on this in human factors. Don Norman has been into errors since Three Mile Island and has several articles out on it -- a popular version is in Psychology Today (last year I think). John Senders has been looking at errors a good deal longer and has also published some interesting work. Finally, a long review of error literature was published by the Nuclear Research Commission (last year again) as a technical report. -------- Although I haven't seen any research devoted specifically to this topic, much work has been done in the area of general-purpose spelling correction. I'm writing a program development system right now (you may have seen my netnews message a few months ago asking about C coding standards), and I will probably implement a simple spelling corrector that utilizes digram analysis or somesuch method. I'm also looking at more subtle errors such as type errors in some so-called strongly-typed languages (such as C). C also lends itself to errors in operator precedence, for example "if (i&FLAGS == 1)" or "if (a=b)". The most useful reference I've seen is the documentation for the "lint" program. It lists lots of common errors for C programs. ------- Norman, Donald A Categorization of Action Slips Psychological Review, 1981, vol. 88, pp 1-15