Multitask Learning for Crash Analysis: A Fine-Tuned LLM Framework Using Twitter Data

Abstract

Highlights: What are the main findings? Demonstrates the effectiveness of a novel multitask learning (MTL) framework utilizing large language models (LLMs) for real-time analysis of road traffic crashes (RTCs) through the integration of social media data. Fine-tuning GPT-2 for language modeling demonstrated that it outperformed baseline models, including GPT-4o mini in zero-shot mode and XGBoost, across various classification and information retrieval tasks. This study benchmarks the performance of the fine-tuned GPT-2 model against these baselines, highlighting its superior performance in these tasks. The study collected and curated a dataset of 26,226 RTC-related tweets from Australia over a year. This dataset extracted fifteen unique features, with six used in classification tasks and nine in information retrieval tasks. Developed an advanced automated labeling system using GPT-3.5, followed by rigorous expert verification to ensure the accuracy and reliability of feature extraction from tweets. The resulting meticulously curated dataset serves as a foundational resource for training and validating subsequent models, establishing a new standard for RTC analysis. What is the implication of the main finding? Offers a transformative approach to traffic safety analytics, providing detailed, timely insights crucial for emergency responders, urban planners, and policymakers. By leveraging cutting-edge AI techniques within an MTL framework, this study demonstrates a transformative approach to real-time RTC analysis, setting the stage for future advancements in the field. The curated dataset generated in this research not only advances traffic safety measures but also serves as a valuable resource for extracting insights, developing models, and conducting further research. This resource provides a solid foundation for future studies aimed at enhancing road safety. Road traffic crashes (RTCs) are a global public health issue, with traditional analysis methods often hindered by delays and incomplete data. Leveraging social media for real-time traffic safety analysis offers a promising alternative, yet effective frameworks for this integration are scarce. This study introduces a novel multitask learning (MTL) framework utilizing large language models (LLMs) to analyze RTC-related tweets from Australia. We collected 26,226 traffic-related tweets from May 2022 to May 2023. Using GPT-3.5, we extracted fifteen distinct features categorized into six classification tasks and nine information retrieval tasks. These features were then used to fine-tune GPT-2 for language modeling, which outperformed baseline models, including GPT-4o mini in zero-shot mode and XGBoost, across most tasks. Unlike traditional single-task classifiers that may miss critical details, our MTL approach simultaneously classifies RTC-related tweets and extracts detailed information in natural language. Our fine-tunedGPT-2 model achieved an average accuracy of 85% across the six classification tasks, surpassing the baseline GPT-4o mini model’s 64% and XGBoost’s 83.5%. In information retrieval tasks, our fine-tuned GPT-2 model achieved a BLEU-4 score of 0.22, a ROUGE-I score of 0.78, and a WER of 0.30, significantly outperforming the baseline GPT-4 mini model’s BLEU-4 score of 0.0674, ROUGE-I score of 0.2992, and WER of 2.0715. These results demonstrate the efficacy of our fine-tuned GPT-2 model in enhancing both classification and information retrieval, offering valuable insights for data-driven decision-making to improve road safety. This study is the first to explicitly apply social media data and LLMs within an MTL framework to enhance traffic safety.