Think Before you Write: QA-Guided Reasoning for Character Descriptions in Books

2026-04-13

AI Summary

This work addresses the challenges of generating accurate character descriptions from novels, which include tracking evolving attributes, synthesizing dispersed evidence, and inferring implicit information. The authors propose a two-stage framework that decouples reasoning from generation: first, a question-answering–guided structured reasoning mechanism produces a faithful reasoning trace grounded in the source text; then, a generative model produces the final character description based on this trace. This approach effectively circumvents interference from the built-in reasoning of large language models and is compatible with both long-context processing and chunked input strategies. Evaluated on the BookWorm and CroSS datasets, the method significantly outperforms strong baselines in terms of faithfulness, informativeness, and grounding in textual evidence.

Technology Category

• Natural Language Processing: Generation
• Knowledge Representation and Reasoning: Reasoning with Beliefs
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Character description generation is an important capability for narrative-focused applications such as summarization, story analysis, and character-driven simulations. However, generating accurate character descriptions from long-form narratives (e.g., novels) is challenging: models must track evolving attributes (e.g., relationships and events), integrate evidence scattered across the text, and infer implicit details. Despite the success of reasoning-enabled LLMs on many benchmarks, we find that for character description generation their performance improves when built-in reasoning is disabled (i.e., an empty reasoning trace). Motivated by this, we propose a training framework that decouples reasoning from generation. Our approach, which can be applied on top of long-context LLMs or chunk-based methods, consists of a reasoning model that produces a structured QA reasoning trace and a generation model that conditions on this trace to produce the final character description. Experiments on two datasets (BookWorm and CroSS) show that QA-guided reasoning improves faithfulness, informativeness, and grounding over strong long-context baselines.

Problem

• character description generation
• long-form narratives
• faithfulness
• reasoning
• implicit inference

Innovation

• QA-guided reasoning
• character description generation
• reasoning-generation decoupling
• long-form narrative understanding
• structured reasoning trace