The University of Sussex

Evaluation of contrasting Prolog trace outputs

Mukesh J. Patel, Benedict du Boulay

We report the effect of differences in format of Prolog trace outputs on Prolog problem solving tasks. Three different trace formats (Spy, TPM*, and EPTB) were tested to check for their relative effectiveness in aiding solution to five different Prolog problems. 43 subjects attempted to solve each problem with each trace (15 problems in total). Analysis of solution times and response data indicates that the EPTB format performed best across all problems. An account for this finding is presented, as is one for a number of interactions between problem type and trace format. These support the general conclusion that not only is format a significant determiner of access to information, but that it can affect users cognitive ability to solve certain sorts of problem associated with Prolog programming.


This paper is not available online