Tree-of-Text: A Tree-based Prompting Framework for Table-to-Text Generation in the Sports Domain

2026-04-29 · Source: Computation and Language · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Natural Language Processing · Depth: Expert, quick

Summary

Tree-of-Text is a novel tree-structured prompting framework designed to generate sports game reports from structured tables, addressing the limitations of traditional model-based and existing prompt-based methods. It guides large language models (LLMs) through a three-stage process: Content Planning, which selects relevant operations and arguments from input tables; Operation Execution, which breaks down large tables into manageable sub-tables; and Content Generation, which merges and rewrites short textual outputs into a cohesive report. This framework aims to mitigate hallucination and improve table comprehension. Experiments demonstrate that Tree-of-Text outperforms existing methods on ShuttleSet+, achieves leading RG and CO metrics on RotoWire-FG, and excels in CS and CO on MLB, while also being approximately 40% faster and more cost-effective than Chain-of-Table.

Key takeaway

For AI Engineers developing table-to-text solutions, especially in data-rich domains like sports, you should consider adopting a structured prompting framework like Tree-of-Text. This approach can significantly improve report generation accuracy and coherence by guiding LLMs through explicit planning and execution stages, potentially reducing development costs and inference time compared to less structured methods like Chain-of-Table.

Key insights

Tree-of-Text uses a tree-structured prompting framework to improve LLM-based table-to-text generation for sports reports.

Principles

• Decompose complex tasks into structured sub-tasks.
• Guide LLMs through explicit content planning and execution.

Method

Tree-of-Text employs a three-stage process: Content Planning (selects operations/arguments), Operation Execution (breaks tables into sub-tables), and Content Generation (merges/rewrites text into reports).

In practice

• Apply tree-structured prompting for complex data-to-text tasks.
• Use sub-table decomposition to manage large inputs.
• Integrate content planning to reduce LLM hallucination.

Topics

• Tree-of-Text
• Table-to-Text Generation
• Prompting Framework
• Large Language Models
• Sports Domain

Best for: AI Engineer, Machine Learning Engineer, Research Scientist, AI Scientist, NLP Engineer, Prompt Engineer

Related on AIssential

• HoWToBench: Holistic Evaluation for LLM's Capability in Human-level Writing using Tree of Writing — Tree-of-Writing (ToW) offers a robust, tree-structured method for evaluating LLM writing, correlating highly with human judgment.
• How Can We Synthesize High-Quality Pretraining Data? A Systematic Study of Prompt Design, Generator Model, and Source Data — Structured pedagogical formats and diverse outputs are key to high-quality synthetic pretraining data, not larger generator models.
• ttda704 at SemEval-2026 Task 6: Structured Chain-of-Thought Prompting for Political Evasion Detection — Structured Chain-of-Thought prompting significantly improves political evasion detection over fine-tuning large language models.
• LEAF: Growing Trees Without Branching for Speech-Aware Large Language Model Post-Training — LEAF is a retrospective tree-based RL method for speech-aware LLM post-training that refines credit assignment.
• Code-on-Graph: Iterative Programmatic Reasoning via Large Language Models on Knowledge Graphs — Code-on-Graph integrates LLMs with KGs using programmatic reasoning, representing schemas as Python classes for flexible, scalable knowledge utilization.
• Derivation Prompting: A Logic-Based Method for Improving Retrieval-Augmented Generation — Derivation Prompting uses logic-inspired rule application to enhance RAG generation, reducing LLM hallucinations and improving interpretability.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computation and Language.