Future work

Future works on the workbench during next steps of the CALIN project should include:

• Optimizations for the full Le Monde corpus (1 million of words)
• Search tools on words, tags or tree patterns using FSA based processing.
• Statistics and frequencies in the corpora.

Further research will try to formalize in our model the principle of coding module introduced in the MATE framework [Dybkjær et al.1998]. Our goal is then to integrate to the workbench functionalities such as:

• Merging several standard one-level annotations of a corpus into our multilevel framework.
• Projecting a particular level of annotation of a multi-level annotated corpus in an existing standard format.

Moreover we want to use the encoding principles proposed here to more complex multimodal annotated corpora resulting from a Wizard of Oz experience (combination of speech and gesture interaction). We also plan to study the uniform realization of the RROM model in XML on the basis of the XML schema proposals.