A central part of law studies in Germany consists of working on legal cases. To
learn the necessary skills, students must exercise with as many sample cases as
possible. But a lot of effort is needed for correction of the students'
solutions. CorAs assists in the correction of solutions to such legal cases.
Often the outline of the solution can already show, if the correct approach has
been chosen and the most important points have been handled. CorAs looks at this
outline and compares it to a sample solution, creating a automatic correction.
In comparison to similar existing programs, the student has to write the whole
solution outline without given bullet points to decide on.
The program displays the student's and the corrector's solutions side by side in
a tree view. Nodes in the learner's solution and the sample are compared,
matching nodes are marked in the same color. The similarity heuristic especially
weighs legal clauses. For each point in the sample, the algorithm decides, if
the learner has covered it well, partly, has a wrong decision or did not mention
it at all.
Correction view: outlines of sample solution (left) and learner's result (right)
After the automatic correction, a human corrector can check the solution and
modify the ratings as well as the matching of nodes in the two solutions.
Finally a feedback document is generated, which explains to the learner, which
parts of the solution were good and which parts could be improved upon.