Analysis and Classification of Programming Exercises by Graph Clustering for Recognition of Model Solutions

Abstract:

Computer programming is a cognitive and formal problem solving process that can involve many possible solutions. Thus, manual evaluation of programming exercises is an onerous task, in particular in the case of numerous exercises and programming classes with many students. Once the assessment is automated, the effort put forth by teachers can be reduced; however, he should consider all possible solutions for each exercise to create model solutions or to train automatic assessment systems. In order to assist teachers in analyzing programming exercise solutions, this paper proposes a strategy based on clustering and LSA (Latent Semantic Analysis) techniques to identify classes of solutions that represent rubrics and automatically sort based on score the majority of the sets of exercise solutions. The results of the first experiments indicate the ability of this strategy to identify solutions classes and to automatically classify the best solutions.

Keywords:

Analysis of Exercises, Clustering, Programming, Rubrics

Citation as:

Marcia G. Oliveira, Howard Roatti, Elias S. Oliveira (2018). "Analysis and Classification of Programming Exercises by Graph Clustering for Recognition of Model Solutions", Computer Communication & Collaboration, Vol. 6, Issue 4, pp. 1-9.