Explanation generation is an important area in human-computer interaction. This includes text planning, organizing dialogues, and user modeling. In text planning, it is important to represent strategies for paragraph-length explanation either in terms of schemata or in terms of rhetorical relations. There exist several attempts that use text planning as a means of generating a text explanation about the structure of physical objects or relations in database.
In this thesis, We proposed a model of text planning which includes a schema for explaning about organizational entities. The schema is consists of two sub-schema, a general schema and a domain specific schema. A general schema is used to select an appropriate explanation structure for a given object. A domain-specific schema fills each sentence fragment with proper data from the pre-defined predicate pool according to the selected structure. In addition, data sets are linked altogether in network configuration in order to prevent data redundancy and to preserve data consistency in generating explanation. Each link in the data set is assigned a value for selecting an appropriate explanation skeleton. The implementation domain is the explanation of the activities in Center for Artificial Intelligence Research at Korea Advanced Institute of Science and Technology. The experiments have shown that the model is effective in generating explanations about laboratories and research areas.
