Ontologically Faithful Generation of Non-Player Character Dialogues

TL;DR

This paper introduces KNUDGE, a new language generation task for creating branching, lore-faithful NPC dialogues in a popular video game, highlighting the challenges of realism and accuracy.

Contribution

It presents a novel dialogue generation task grounded in a real game environment, emphasizing complex branching structures and lore fidelity, with initial neural model results.

Findings

Neural models perform competently on KNUDGE tasks.

Branching dialogue trees are more challenging than linear dialogues.

Future work needed to improve realism and game-quality of generated dialogues.

Abstract

We introduce a language generation task grounded in a popular video game environment. KNUDGE (KNowledge Constrained User-NPC Dialogue GEneration) requires models to produce trees of dialogue between video game characters that accurately reflect quest and entity specifications stated in natural language. KNUDGE is constructed from side quest dialogues drawn directly from game data of Obsidian Entertainment's The Outer Worlds, leading to real-world complexities in generation: (1) dialogues are branching trees as opposed to linear chains of utterances; (2) utterances must remain faithful to the game lore -- character personas, backstories, and entity relationships; and (3) a dialogue must accurately reveal new quest details to the human player. We report results for a set of neural generation models using supervised and in-context learning techniques; we find competent performance but room for future work addressing the challenges of creating realistic, game-quality dialogues.