Domain-Independent Automatic Generation of Descriptive Texts for Time-Series Data

2024-09-25

IEEE International Conference on Acoustics, Speech, and Signal Processing

AI Summary

Time-series data often lack descriptive textual annotations, hindering the training of text generation models. Method: This paper introduces a novel forward-and-backward dual-path pairing construction framework, pioneering a backward generation paradigm: starting from human-crafted rule-based textual descriptions, it synthesizes corresponding time-series data in reverse. Based on this paradigm, we construct TACO—the first cross-domain time-series–text paired dataset. We further propose a contrastive learning–based text generation framework that jointly models abstract time-series features and incorporates backward rule guidance. Results: Experiments demonstrate zero-shot textual generation on unseen domains, significantly improving cross-domain generalization and generated text quality. Moreover, the approach enhances interpretability of time-series understanding and boosts human–machine interaction efficiency.

Technology Category

• Natural Language Processing: Generation
• Machine Learning: Time-Series/Data Streams
• Application Domains: Humanities & Computational Social Science

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Web data generation and simulation
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Due to scarcity of time-series data annotated with descriptive texts, training a model to generate descriptive texts for time-series data is challenging. In this study, we propose a method to systematically generate domain-independent descriptive texts from time-series data. We identify two distinct approaches for creating pairs of time-series data and descriptive texts: the forward approach and the backward approach. By implementing the novel backward approach, we create the Temporal Automated Captions for Observations (TACO) dataset. Experimental results demonstrate that a contrastive learning based model trained using the TACO dataset is capable of generating descriptive texts for time-series data in novel domains.

Problem

• Generates descriptive texts for time-series data automatically
• Overcomes scarcity of annotated time-series text data
• Proposes domain-independent method using contrastive learning

Innovation

• Domain-independent text generation from time-series data
• Novel backward approach for dataset creation
• Contrastive learning model trained on TACO dataset