Writing Like the Best: Exemplar-Based Expository Text Generation

2025-05-24

AI Summary

To address key bottlenecks in analogical expository text generation—including strong exemplar dependency, poor topic adaptation, and degraded long-text coherence—this paper proposes an adaptive imitation paradigm and introduces the iterative “Planning–Adaptation” (RePA) framework. RePA integrates an LLM-driven fine-grained planning–adaptation mechanism, a dual-memory architecture (comprising input clarification memory and output coherence memory), and iterative segmented generation. We further propose three novel evaluation metrics: imitation fidelity, topic adaptability, and adaptive imitation capability. Extensive experiments on three custom-built, multi-domain datasets demonstrate that our method significantly outperforms baselines across factual accuracy, logical consistency, and topic relevance. The framework offers a scalable, resource-efficient solution for high-fidelity expository text generation, particularly beneficial in low-resource settings.

Technology Category

• Natural Language Processing: Generation
• Cognitive Modeling & Cognitive Systems: Analogy
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

We introduce the Exemplar-Based Expository Text Generation task, aiming to generate an expository text on a new topic using an exemplar on a similar topic. Current methods fall short due to their reliance on extensive exemplar data, difficulty in adapting topic-specific content, and issues with long-text coherence. To address these challenges, we propose the concept of Adaptive Imitation and present a novel Recurrent Plan-then-Adapt (RePA) framework. RePA leverages large language models (LLMs) for effective adaptive imitation through a fine-grained plan-then-adapt process. RePA also enables recurrent segment-by-segment imitation, supported by two memory structures that enhance input clarity and output coherence. We also develop task-specific evaluation metrics--imitativeness, adaptiveness, and adaptive-imitativeness--using LLMs as evaluators. Experimental results across our collected three diverse datasets demonstrate that RePA surpasses existing baselines in producing factual, consistent, and relevant texts for this task.

Problem

• Generating expository texts on new topics using similar exemplars
• Overcoming reliance on extensive exemplar data and topic adaptation issues
• Improving long-text coherence in exemplar-based text generation

Innovation

• Adaptive Imitation concept for text generation
• Recurrent Plan-then-Adapt framework
• LLM-based evaluation metrics