Extending Narrative Planning Domains with Linguistic Resources

Abstract

Interactive Narrative is an emerging application of Planning technologies in which plan-based narrative structures are used to control virtual character behaviours. Despite popularity in research prototypes, more widespread adoption has been hindered by the difficulty of authoring planning domains, even in interdisciplinary teams, as such applications require domain models which are robust to dynamic environments as well as being capable of generating a diversity of solutions. To reduce the burden of creating models for such systems we explored automated extension of partially developed planning domains in order to address the issues of robustness and diversity. In the paper we introduce a novel algorithm, THYPE, which proposes additional types of virtual characters and narrative objects with which to extend a domain model. These extensions increase the robustness of domain models by enlarging the range of type of characters and objects with which to enact planned behaviours. We see THYPE being used in an off-line process, within an authoring support tool for the narrative planning domain. Hence it is important that the proposed extensions have human-readable semantic labels. In the paper we embed THYPE within a modular and extensible framework that allows the combination of multiple off-line generated extensions. In the paper we present results of a user study that shows that THYPE suggestions are plausible. We also empirically demonstrate the ability of THYPE extensions to increase the robustness of the models, measured as likelihood of plan continuation through to an original goal in the space of changes in the execution environment at run-time. Further, we demonstrate the modular nature of our extension framework by combining THYPE with another extension approach dedicated to recovery from plan failure. We present results which show that enhanced performance results from the combination of multiple extensions.

Publication
In AAMAS 2020
Avatar
João F. Ferreira
Computer Scientist

My research interests include software reliability, software verification, and formal methods applied to software engineering. I am also interested in interactive storytelling. For more details, see some of my projects or my selected (or recent) publications. More posts are available in my blog. Follow me on Twitter or add me on LinkedIn.