Application of Chinese Natural Language Generation in Semantic Web

RDF is the representation of the Semantic Web. When querying RDF documents, the result is a sub-graph of RDF data model or a number of triple statements. In this paper, we apply natural language generation technique to render the result into multi-sentential text for human comprehension. We investigate the effect of discourse segmentation on the generation of anaphora and punctuation marks in Chinese. We first describe the construction of hierarchical discourse structure of message content by employing the planning concept in the field of artificial intelligence. Based on this discourse structure, we propose to segment the discourse in terms of the levels of rhetorical nodes in the hierarchy of message structure. The implementation result shows that the segmentation scheme is promising.