From Metrics to Meaning: Rule-Grounded LLM Explanations for Data Literacy in the Case of Youth Football

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Sports Technology & Analytics · Depth: Expert, quick

Summary

A rule-grounded data-to-text framework has been developed to enhance data literacy in youth football by providing concise, stakeholder-specific summaries of training sessions. This framework addresses the challenge of interpreting wearable technology training metrics for young athletes, parents, and coaches. It operates by mapping duration-normalized indicators to structured facts concerning session profile, internal intensity, speed exposure, and movement dynamics. These facts are then verbalized by a large language model tailored for different audiences. The research compared direct generation from raw metrics, generation from rule-derived facts, and an augmented "ENRICHED" configuration, which combines validated facts with raw metrics and explicit thresholds, utilizing LoRA-adapted open-weight models. Developed with 122 anonymized U15 player-session records and evaluated on ten held-out sessions, the framework demonstrated that rule grounding significantly improves reliability and audience adaptation, particularly by mitigating unsupported interpretations. Expert evaluation by physical education teachers confirmed the ENRICHED setting's player-facing explanations were accurate, comprehensible, and practically useful.

Key takeaway

For sports analytics developers building data interpretation tools, you should integrate rule-grounded LLM explanations to enhance reliability and audience adaptation. This approach, particularly the ENRICHED configuration, significantly reduces unsupported interpretations of complex training metrics. Consider adapting open-weight models with LoRA to generate accurate, comprehensible, and practically useful summaries for diverse stakeholders like athletes, parents, and coaches, thereby improving data literacy and trust in your systems.

Key insights

Rule grounding LLM explanations improves reliability and audience adaptation for complex data interpretation.

Principles

• Contextual explanations are crucial for data literacy.
• Rule grounding enhances LLM reliability.
• Stakeholder-specific summaries improve utility.

Method

A data-to-text framework maps duration-normalized training metrics to structured facts (session profile, intensity, speed, movement). These facts are then verbalized by an LLM, optionally augmented with raw metrics and thresholds (ENRICHED).

In practice

• Apply rule grounding to sports analytics.
• Use LoRA for LLM adaptation.
• Tailor explanations for specific user roles.

Topics

• Data Literacy
• Large Language Models
• Sports Analytics
• Rule-Grounded AI
• Wearable Technology
• Explainable AI

Best for: AI Scientist, Research Scientist, Domain Expert

Related on AIssential

• Using Large Language Models and Knowledge Graphs to Improve the Interpretability of Machine Learning Models in Manufacturing — Combining LLMs and KGs can generate user-friendly, context-aware explanations for ML model predictions.
• ReviewGrounder: Improving Review Substantiveness with Rubric-Guided, Tool-Integrated Agents — Integrating rubrics and contextual grounding significantly enhances LLM-generated peer review quality and substantiveness.
• LLM Predictive Scoring and Validation: Inferring Experience Ratings from Unstructured Text — LLMs can directionally predict experience ratings from unstructured text, revealing distinct evaluative constructs.
• Boundary-Aware Context Grounding for A Low-Channel EEG Agent — Separating EEG processing from LLM interpretation via hardware-aware grounding prevents unsupported conclusions.
• Part 2: Turning My Writing into a Training Set — A method called "grounded generation" creates personalized fine-tuning datasets by constraining LLMs to source documents, ensuring authentic voice and content.
• Reasoning as Pattern Matching: Shared Mechanisms in Human and LLM Everyday Reasoning — Human and LLM everyday reasoning aligns with content-sensitive pattern-matching, challenging abstract world model assumptions.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.