A Domain-Adapted Pipeline for Structured Information Extraction from Police Incident Announcements on Social Media

2025-12-18

AI Summary

Structural information extraction from informal, noisy, and heterogeneous Weibo public security alert texts remains challenging due to linguistic irregularities and source heterogeneity. Method: This paper proposes a synergistic framework integrating prompt engineering tailored to policing contexts with Low-Rank Adaptation (LoRA) fine-tuning, built upon the Qwen2.5-7B foundation model. It leverages domain-specific prompts, LoRA-based parameter-efficient adaptation, and a high-quality human-annotated dataset (4,933 instances) to perform end-to-end joint extraction of 15 critical fields—including location, incident characteristics, and casualty assessment. Results: The framework achieves 98.36% accuracy in fatality determination, and exact-match rates of 95.31% for fatality counts and 95.54% for provincial-level locations—substantially outperforming baseline and instruction-tuned models. To our knowledge, this is the first work to deeply integrate prompt engineering with LoRA for structured information extraction from Chinese policing texts, demonstrating both strong domain adaptability and state-of-the-art precision.

Technology Category

• Natural Language Processing: Information Extraction
• Data Mining & Knowledge Management: Intelligent Query Processing
• Search and Optimization: Algorithm Configuration

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Structured information extraction from police incident announcements is crucial for timely and accurate data processing, yet presents considerable challenges due to the variability and informal nature of textual sources such as social media posts. To address these challenges, we developed a domain-adapted extraction pipeline that leverages targeted prompt engineering with parameter-efficient fine-tuning of the Qwen2.5-7B model using Low-Rank Adaptation (LoRA). This approach enables the model to handle noisy, heterogeneous text while reliably extracting 15 key fields, including location, event characteristics, and impact assessment, from a high-quality, manually annotated dataset of 4,933 instances derived from 27,822 police briefing posts on Chinese Weibo (2019-2020). Experimental results demonstrated that LoRA-based fine-tuning significantly improved performance over both the base and instruction-tuned models, achieving an accuracy exceeding 98.36% for mortality detection and Exact Match Rates of 95.31% for fatality counts and 95.54% for province-level location extraction. The proposed pipeline thus provides a validated and efficient solution for multi-task structured information extraction in specialized domains, offering a practical framework for transforming unstructured text into reliable structured data in social science research.

Problem

• Extract structured information from noisy police social media announcements
• Handle informal text variability in Chinese Weibo police incident posts
• Develop domain-adapted pipeline for multi-task key field extraction

Innovation

• Domain-adapted pipeline with targeted prompt engineering
• LoRA fine-tuning of Qwen2.5-7B for noisy text extraction
• Extracts 15 key fields from police social media posts