Character Beliefs in Story Generation

Abstract

Character beliefs play a central role in many narratives, but are often not represented in planning-based author-centered narrative generation systems, or only represented in an ad hoc way. In this paper we will discuss how actions in Dynamic Epistemic Logic that affect the story world as well as the characters' beliefs, can be used in the context of narrative generation. By using this logical foundation we can ensure that the actors' beliefs are maintained in a logically consistent way, but we will also discuss how our system supports writing these complex logical operations in a concise way. We will also show how our system, which has limited planning capabilities, can come up with simple stories that cause characters to have beliefs desired by the author, and how our approach can be integrated with other work in the field.

Murder Mysteries: The White Whale of Narrative Generation?

“…At this point, we also find it important to note that we in no way hold the limitations of their systems, or the lack of further progress, against any of these authors. On the contrary, we believe this just aids to illustrate our point that the problem is much more challenging that it might appear at first glance, and we ourselves have presented yet another approach based on logic to model belief updates that can be added to the list of projects awaiting further work (Eger and Martens 2017a). It was this work exactly, and our investigation into solutions other authors may have come up with, that lead us to write the present article in the first place.…”

Sabre: A Narrative Planner Supporting Intention and Deep Theory of Mind

“…Multi-agent systems like Thespian (Si and Marsella 2014) and Talk of the Town (Ryan et al 2015) reason about theory of mind but have limited centralized planning, so they would need to run many times until they happen to achieve the author's goal. Ostari (Eger and Martens 2017) is centralized with intentions and beliefs but allows true uncertainty, which dramatically increases the size of its search space. As a test, the smallest example from the Lovers domain (described below) was implemented and tested in Ostari, but it quickly ran out of memory before finding any solutions.…”

“…Previously we defined a state space with these features (Shirvani, Ware, and Farrell 2017) but focused on validating the knowledge representation and did not provide a planning algorithm or measuring performance. Ostari (Eger and Martens 2017) has these features but models true uncertainty via Dynamic Epistemic Logic. The high cost of modeling all doxastically accessible possible worlds limits the scope of problems it can solve.…”

Generating Explorable Narrative Spaces with Answer Set Programming

“…To address these goals, we present a narrative planning engine implemented with Answer Set Programming (ASP), and show how to layer narrative constraints and formulate use-specific interactions expressed as ASP clauses. We use the expressiveness afforded by ASP constraints to implement notions of intention, conflict, and belief proposed as key aspects of character believability by previous work (Riedl and Young 2010;Ware and Young 2011;Wadsley and Ryan 2013;Eger and Martens 2017;Shirvani, Farrell, and Ware 2018). We then show how global narrative constraints, such as story grammars and plot structures, can be incorporated.…”