ducut91/traffic-dispute-mediation-cases

🚗 Korean Traffic Dispute Mediation Cases Dataset

This dataset provides a comprehensive collection of 226 traffic accident fault ratio dispute cases from Korean Automobile Insurance Association (KNIA), focusing on traffic accidents, fault determination, and mediation outcomes. It is suitable for traffic-legal NLP, fault analysis, mediation outcome prediction, and automotive AI applications.

---

📦 Dataset Summary

• Total Records: 226
• Language: Korean (ko)
• Fields: 9 columns with English names
• Source: Korea Non-life Insurance Association (KNIA) - Automobile Accident Fault Ratio Dispute Review Committee
• Time Period: 2016-2019
• License: This dataset is based on public mediation case data and is freely available for research, educational, and commercial purposes.

---

🧾 Column Descriptions

Column Name	Description
department	Department handling the dispute (자동차사고과실비율분쟁심의위원회)
id	Unique case identifier (format: YYYY-XXXXXX)
title	Brief description of the traffic accident type and circumstances
url	Link to the original case document on the official KNIA website
content	Detailed case information including accident details, claims, evidence, and decision
summary	Brief summary of the accident and final fault determination
facts	Key factual circumstances of the traffic accident
issue	Primary legal and factual issues in dispute
outcome	Final fault ratio determination and reasoning

---

📊 Missing Value Statistics

Column Name	Missing Count	Missing Ratio (%)
department	0	0.00%
id	0	0.00%
title	0	0.00%
url	0	0.00%
content	0	0.00%
summary	0	0.00%
facts	0	0.00%
issue	0	0.00%
outcome	0	0.00%

✅ Perfect Data Completeness: No missing values in any column (100% complete).

---

🚦 Traffic Accident Type Distribution

The dataset covers various traffic accident scenarios with the following distribution:

Accident Type	Cases	Percentage	Description
직진사고 (Straight-line Accidents)	129	57.1%	Collisions between vehicles going straight
기타 (Other Types)	74	32.7%	Various other traffic accident scenarios
좌회전사고 (Left Turn Accidents)	12	5.3%	Accidents involving left-turning vehicles
우회전사고 (Right Turn Accidents)	10	4.4%	Accidents involving right-turning vehicles
후진사고 (Reverse Accidents)	1	0.4%	Accidents involving reversing vehicles

---

📅 Temporal Distribution

Year	Cases	Percentage	Description
2019	113	50.0%	Most recent cases with comprehensive documentation
2018	84	37.2%	Significant portion of mediation cases
2017	26	11.5%	Earlier period cases
2016	3	1.3%	Limited early dataset cases

Peak Period: 2018-2019 accounts for 87.2% of all cases, indicating comprehensive data collection during this period.

---

🛠️ Data Collection Notes

• Source: Korea Non-life Insurance Association (KNIA)
• Original Document: 자동차사고 과실비율분쟁 심의사례집 (2020) PDF
• Collection Method: PDF processing and data extraction from official case compilation
• Data Quality: All cases include complete documentation of accident circumstances, legal arguments, evidence analysis, and fault ratio decisions

---

💡 Suggested Use Cases

• 🚗 Traffic accident analysis – Analyze patterns in vehicular collision cases
• 🤖 Fault prediction models – Train models to predict fault ratios in traffic disputes
• 📊 Insurance risk assessment – Identify common accident scenarios and liability patterns
• 💬 Legal Q&A systems – Build specialized chatbots for traffic accident disputes
• 📈 Traffic safety research – Study trends in traffic accident causes and prevention
• 🧠 Legal document summarization – Train models on traffic-legal text processing
• 🏛️ Comparative legal analysis – Compare Korean traffic law applications with other jurisdictions

---

📊 Key Statistics

• Total cases: 226 traffic accident dispute mediation cases
• Case ID range: 2016-051544 to 2019-XXXXXX (chronological numbering)
• Average content length: ~1,590 characters
• Content length range: 1,122 to 2,139 characters
• Average summary length: ~83 characters
• Most common accident type: 직진사고 (Straight-line accidents) - 57.1%
• Peak years: 2018-2019 (87.2% of all cases)
• Data completeness: 100% (no missing values)

---

📜 License & Legal Considerations

This dataset is based on arbitration decisions from the "자동차사고 과실비율분쟁 심의사례집 (2020)" published by the Korea Non-life Insurance Association (KNIA).

Legal Framework:

• Copyright Law Article 7(3): Under Korean Copyright Act, decisions by arbitration procedures and "similar procedures" are not protected by copyright
• Quasi-Judicial Process: The automobile accident fault ratio dispute review constitutes a quasi-judicial arbitration procedure similar to administrative arbitration
• Public Domain Status: The original arbitration decisions fall under public domain as dispute resolution decisions
• Open License: Released under CC0 (Public Domain) for unrestricted use

Privacy Protection: All cases are anonymized with personal information removed. Vehicle owners, drivers, and insurance company identities are protected according to Korean privacy laws.

✅ Public Domain: This dataset can be freely used for any purpose without restrictions.

---

🎯 Why We Open-Sourced This

We believe that global large language models (LLMs) should better understand and reflect the legal system of South Korea, just as they do for other major jurisdictions.

By releasing this dataset publicly, we hope to:

• Promote inclusion of Korean law in multilingual legal AI systems
• Encourage international research in Korean legal language processing
• Lower the barrier to entry for developers working on Korean legal services

🇰🇷 A legally informed model is a better model — for everyone.

---

📅 Changelog

Date	Changes Made
2025-08-01	Initial dataset release with 226 traffic dispute mediation cases