Query-Conditioned Graph Retrieval for Contextualized LLM Reasoning in Personalized Wearable Data

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Lu, Zhenyu, Abbasian, Mahyar, Rahmani, Amir M.
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866917509328797696
author Lu, Zhenyu
Abbasian, Mahyar
Rahmani, Amir M.
author_facet Lu, Zhenyu
Abbasian, Mahyar
Rahmani, Amir M.
contents Large language models (LLMs) are increasingly applied to analyzing wearable sensing data, which are long-term, multimodal, and highly personalized. A key challenge is context selection: providing insufficient context limits reasoning, while including all available data leads to inefficiency and degraded generation quality. We propose Wearable As Graph (WAG), a graph-based context retrieval framework that enables query-adaptive reasoning over wearable data with LLMs. WAG organizes wearable metrics and user-specific signals into a personalized knowledge graph, and retrieves a query-conditioned subgraph to support downstream generation. The retrieval process integrates global relationships, capturing prior knowledge and population- and individual-level patterns via hierarchical Bayesian modeling, with local relationships that reflect short-term signal deviations. A query openness signal further controls retrieval breadth. We evaluate WAG on over 10,000 data-grounded queries from real-world wearable datasets. Across LLM-based and human evaluations, WAG achieves an approximately 70% win rate over baseline and standard RAG methods, demonstrating the effectiveness of structured, query-adaptive context retrieval for LLM-driven analysis of wearable data.
format Preprint
id arxiv_https___arxiv_org_abs_2605_18763
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Query-Conditioned Graph Retrieval for Contextualized LLM Reasoning in Personalized Wearable Data
Lu, Zhenyu
Abbasian, Mahyar
Rahmani, Amir M.
Information Retrieval
Artificial Intelligence
Large language models (LLMs) are increasingly applied to analyzing wearable sensing data, which are long-term, multimodal, and highly personalized. A key challenge is context selection: providing insufficient context limits reasoning, while including all available data leads to inefficiency and degraded generation quality. We propose Wearable As Graph (WAG), a graph-based context retrieval framework that enables query-adaptive reasoning over wearable data with LLMs. WAG organizes wearable metrics and user-specific signals into a personalized knowledge graph, and retrieves a query-conditioned subgraph to support downstream generation. The retrieval process integrates global relationships, capturing prior knowledge and population- and individual-level patterns via hierarchical Bayesian modeling, with local relationships that reflect short-term signal deviations. A query openness signal further controls retrieval breadth. We evaluate WAG on over 10,000 data-grounded queries from real-world wearable datasets. Across LLM-based and human evaluations, WAG achieves an approximately 70% win rate over baseline and standard RAG methods, demonstrating the effectiveness of structured, query-adaptive context retrieval for LLM-driven analysis of wearable data.
title Query-Conditioned Graph Retrieval for Contextualized LLM Reasoning in Personalized Wearable Data
topic Information Retrieval
Artificial Intelligence
url https://arxiv.org/abs/2605.18763