Empirical Methods in Discourse: Interpretation & Generation
Papers from the AAAI Spring Symposium
Johanna Moore & Marilyn Walker,Cochairs
Computational theories of communicative action provide the foundation for the design of a wide range of AI systems (such as intelligent tutors, believable agents, intelligent software agents like softbots, and so forth. To date, much work in computational theories of discourse has focused on specifying the mechanisms underlying a particular discourse phenomenon. However developing the robust, broad coverage, theories of discourse that are needed in today's systems requires an empirical basis. Moreover, there are no shared methods, tools or resources for the discourse community.
This symposium aims to investigate empirical methods that can be used in the development and evaluation of computational theories of discourse, and develop a set of shared resources for the computational discourse community. Symposium presentations will address the following issues:
- Corpus-Based methods applied to theories of discourse.
- Evaluation of dialogue modules in implemented systems.
- Simulation tools for developing and evaluating theories of discourse.
- Coding schemes for the quantitative study of discourse phenomena.
- Tools that support (semi-)automatic or empirical studies of discourse phenomena.
- Applications and extensions of methods used in traditionally empirical disciplines (such as, psychology or sociolinguistics) to computational theories of discourse.
- Empirical analyses that distinguish between claims made by different discourse theories.