Among other things, it allows a user to create domains and problems using tarski objects, with all the necessary functions such as types, predicates, actions, and requirements and the ability to read and write to. The program was written using a custom library called tarski, which enables various PDDL functionality. By merging the two problems we were able to test student submissions to evaluate how correct their assignments were. We considered two planning problems, one being the student’s submission and the other being the reference solution, and assessed how equivalent they were under the lens of regular bisimulation. Bisimiulation is the study of alignment between two dynamical systems. The decision to automate the grading process was motivated by the idea of regular bisimulation to assess if the problems were isomorphic. Previously, the PDDL Assignment had been tedious to mark by the teaching team, as many valid solutions may exist when mapping a situation into the PDDL format. This problem had two aspects: automatically generating correct KenKen boards, given certain requirements as inputs, and generating meaningful feedback for common student errors.
The secondary focus of the project was to automate the grading of the Constraint Satisfaction Problem assignment, which revolves around solving KenKen boards. This enables the teaching team to automatically assess whether a student’s submission is equivalent to the correct solution template. To identify when two problem and domain pairs are isomorphic. The primary focus of the project was the automation of the PDDL Assignment grading, by devising a script using the Python tarski library Several assignments in the CISC352 Introduction to AI course are very time-consuming to mark by the teaching team, and the overall purpose of this project was to automate this process. Both the PDDL and CSP components were run on student submissions from the CISC352 course and effectively reduced marking time for the teaching team. The automation process for this assignment involved automatically generating KenKen boards and outputting meaningful feedback for each student submission. The second component focused on the constraint satisfaction problem (CSP) assignment and involved KenKen boards. The result of this process indicates whether a student’s submission aligns with the reference solution, providing a quick and efficient way to find errors within a student’s submission. This involved creating a script that merged two domain-problem pairs to check for isomorphism. The primary component focused on the grading of the PDDL assignment. This project consisted of two components with the goal of automating the grading process of CISC352 assignments.
0 kommentar(er)
