From discourse analysis to answering design questions”,Proceedings

Abstract. This paper discusses an approach of modelling design rational expressed in natural language sentences into a discourse model. The discourse model is used to classify captured rationale into discourse categories by taking into account semantic and pragmatic relationships between two design elements. It is expected that accessibility and reusability of captured rationale is improved since retrieval is supported within discourse contexts. A small dataset was collected to test whether selected discourse relations are extractable and whether a machine learning algorithm can generate rules under which appropriate relations can be automatically marked.