DiffNator: Generating Structured Explanations of Time-Series Differences

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Dohi, Kota, Nishida, Tomoya, Purohit, Harsh, Endo, Takashi, Kawaguchi, Yohei
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866908556686524416
author Dohi, Kota
Nishida, Tomoya
Purohit, Harsh
Endo, Takashi
Kawaguchi, Yohei
author_facet Dohi, Kota
Nishida, Tomoya
Purohit, Harsh
Endo, Takashi
Kawaguchi, Yohei
contents In many IoT applications, the central interest lies not in individual sensor signals but in their differences, yet interpreting such differences requires expert knowledge. We propose DiffNator, a framework for structured explanations of differences between two time series. We first design a JSON schema that captures the essential properties of such differences. Using the Time-series Observations of Real-world IoT (TORI) dataset, we generate paired sequences and train a model that combine a time-series encoder with a frozen LLM to output JSON-formatted explanations. Experimental results show that DiffNator generates accurate difference explanations and substantially outperforms both a visual question answering (VQA) baseline and a retrieval method using a pre-trained time-series encoder.
format Preprint
id arxiv_https___arxiv_org_abs_2509_20007
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DiffNator: Generating Structured Explanations of Time-Series Differences
Dohi, Kota
Nishida, Tomoya
Purohit, Harsh
Endo, Takashi
Kawaguchi, Yohei
Computation and Language
In many IoT applications, the central interest lies not in individual sensor signals but in their differences, yet interpreting such differences requires expert knowledge. We propose DiffNator, a framework for structured explanations of differences between two time series. We first design a JSON schema that captures the essential properties of such differences. Using the Time-series Observations of Real-world IoT (TORI) dataset, we generate paired sequences and train a model that combine a time-series encoder with a frozen LLM to output JSON-formatted explanations. Experimental results show that DiffNator generates accurate difference explanations and substantially outperforms both a visual question answering (VQA) baseline and a retrieval method using a pre-trained time-series encoder.
title DiffNator: Generating Structured Explanations of Time-Series Differences
topic Computation and Language
url https://arxiv.org/abs/2509.20007