A Way for Using the Verb Conceptual Dictionary in an Integrated Narrative Generation System: the Use of Co-occurrence Information on Verb Concepts
Takashi Ogata, Jumpei Ono · Journal of Robotics Networking and Artificial Life · 2016
This study presents a method of selecting verb concepts based on analysis of co-occurrence information of verbs in modern Japanese novels collected from the Japanese digital library, "Aozora Bunko".We plan to incorporate the proposed method into an integrated narrative generation system (INGS) that synthesizes various mechanisms (story generation, sentence generation, and so on) for narrative generation.We are currently developing such a system.In this study, we provide an overview of INGS and examine mechanisms related to the proposed method.In a verb conceptual dictionary that we have developed, as verb concepts of a variety of difficulty (comprehensibility) are mixed, the objective in this study is to adjust difficulty of verb concepts by using co-occurrence information on verb concepts.Our hypothesis is that the original concepts of co-occurrence and the resulting concepts have a proportional relationship.We investigate this hypothesis by incorporating proposed mechanism into an INGS.