++ Project proposals for the practical exercise hours
Courseware: Extend one of the demo exercises with additional items. Give a justification why you think it is important to have these cases included.
Courseware: Develop an instance-based tutoring system for a learning problem in a language of your choice (preferably your mother tongue). Discuss the application area you have chosen with respect to
the difficulties it (probably) incurs for a second language learner,
the inherent regularities of language, which could be described in a generalised manner,
the chances to cover all positive solutions with a finite collection of anticipated ones
the chances to cover the most relevant student errors (and their combinations!) with a finite set of anticipated ones.
Software: Extend the flexible matching algorithm with a better error description mechanism and implement an error explanation component, which produces natural language utterances.
Software: Analyse the given web server for missing functionality. Design and implement a corresponding alternative.
Lingware: Develop a constraint-based diagnosis for a learning problem in a language of your choice. Discuss the chances for making your solution water-proof.
Software: Extend the constraint-based diagnosis with additional co-occurrence constraints (e.g. implication, exclusive or, single obligatory value assignments) and apply them in a specifically designed exercise.
Software: Implement an explanation component for the constraint-based diagnosis, which produces natural language utterances.
Software: Implement a web interface for the constraint-based diagnosis.
Lingware: Implement a DCG which is able to produce fill-in-the-gap prompts for a particular exercise scenario. Explain the degree of freedom for the student and discuss the suitability of different diagnostic approaches in this scenario.
Lingware/Software: Implement a DCG which
is able to parse ill-formed sentences by means of error-rules and
generates error descriptions for them.
Lingware/Software: Implement a DCG which uses weak feature unification for error diagnosis.