AutoSpecNER: A Fine-Grained Named Entity Recognition Dataset for Vehicle Specification Extraction

2026-06-23

AI Summary

This work addresses the scarcity of fine-grained named entity recognition (NER) resources for extracting detailed vehicle specifications from advertisements, which are rich in such information. To bridge this gap, the authors introduce AutoSpecNER, the first high-quality, expert-annotated NER dataset for vehicle specifications, encompassing 15 distinct entity types. Annotation quality is rigorously ensured through inter-annotator agreement analysis. Comprehensive evaluation demonstrates that a fine-tuned DeBERTa model substantially outperforms both rule-based approaches (micro-F1: 90.0% vs. 43.0%) and existing large language models (77.8%), establishing a strong benchmark and an effective solution for structured information extraction in the automotive domain.

Abstract

Vehicle advertisements contain rich specification information, but automotive NER resources remain limited. We introduce AutoSpecNER, an expert-annotated dataset for fine-grained entity recognition in vehicle listings. The dataset includes 659 advertisements from a popular car-selling website, with over 10,000 entities annotated across 15 categories, including MODEL, ENGINE_SPEC, and BATTERY_CAPACITY. Annotation quality was validated through inter-annotator agreement, achieving an average score of 91.5%. We benchmark rule-based extraction, fine-tuned transformer encoders, and large language models. DeBERTa achieves the best performance with a 90% micro-F1 score, outperforming the rule-based baseline (43%) and the strongest large language model (77.8%).

Problem

• Named Entity Recognition
• Vehicle Specification
• Fine-Grained Entity Recognition
• Automotive NER
• Entity Extraction

Innovation

• fine-grained NER
• vehicle specification extraction
• expert-annotated dataset
• DeBERTa
• inter-annotator agreement