The math use case remains as it was, except that the use case may assume that premises requiring encyclopedic knowledge needed to frame word problems are given. Assuming that the math scenario will be embedded in the semantic wiki, the background premises may be given by the author of the problem in the facts database where the problems are formulated.
The mathematics use cases involve a problem author, a student and a teacher. The usability of the scenario is tested with realistic subjects playing each of these roles and the evaluation collected with a questionnaire and/or a journal. In addition, we should try estimate the savings from the system when scaled up to a larger use base and variety of languages, since these are the novelties in the MOLTO solution.
WP 6 has developed a treebank based method for doing regression testing on the translations produced by the math grammar. A treebank entry consists of:
A Changeset has:
A defect is a difference between the actual linearization of an entry and the sample in the last changeset.
The procdure is as follows.
See http://www.molto-project.eu/wiki/living-deliverables/d61-simple-drill-grammar-library/5-testing for further discussion.