Next: Evaluation of Writers' Aids
Up: Requirements Analysis for Linguistic
Previous: Generic requirements models: adaptation
It should be clear from this document that much work remains to be
done in the area of requirements analysis and user profiling for LE\
evaluation. A few points can be pulled out for particular attention.
- More work needs to be done on requirements analysis guidelines
for LE evaluation, both in the design and accumulation of
libraries of task, setup, and requirements elements, and in the
development of a set of structured knowledge acquisition steps for
- Insights, tools, and procedures from many of the research areas
in the requirements engineering community (such as traceability of
requirements, viewpoint analysis, and validation of requirements)
could be made applicable to LE evaluation with a bit of work.
- Requirements elements could be reused in an extended
Parameterisable Test Bed, in the style of the Generic Task Libraries
of the KADS modelling system, if they can be suitably represented.
Aspects of the requirements analysis process that could be supported
by such a tool would include the development of setup and scenario
descriptions from pre-existing elements, and the recording of
justifications/validation information for requirements
transformations. Recording the process of requirements analysis
serves to legitimate an evaluation in the same way as recording the
methods used to measure system performance.
- Knowledge Acquisition techniques used for Knowledge Based Systems
could aid the systematic development of specifications for LE\
systems. Automated toolkits for KA could be adapted to support
requirements gathering for evaluation.
- Specifically linguistic modelling based on analysis of realistic
corpora needs to be integrated into any such scheme.
Characterisation of corpora in such a way as to support the
definition of modular, hierarchical requirements elements is an