BROWSE
Detail
Conference Paper
Classification of Police Reports and Non-Police Reports with Data Length Normalization Learning
Cited 0 time in 
Share 
Share
- Authors
- Hyunho Park, Eungyeol Lee, Sungwon Byon, Eunjung Kwon, Minjung Lee, Eui-Suk Jung
- Issue Date
- 2024-10
- Citation
- International Conference on Information and Communication Technology Convergence (ICTC) 2024, pp.262-264
- Publisher
- IEEE
- Language
- English
- Type
- Conference Paper
- Abstract
- The classification of police reports and non-police reports is important when emergency police calls (e.g., 112 calls in South Korea) are received at a police station. Non-urgent calls should be promptly classified as non-police reports to reduce the burden on police operations. When conducting training for the classification of police and non-police reports, there is an issue of misclassification due to the differing lengths of training text data between police reports and non-police reports. This paper proposes data length normalization learning (DLNL), which normalizes data for training and learns the normalized data, to improve the classification of police reports and non-police reports. This paper also explains the performance of the improved classification with an F1-score of 0.99. The enhanced classification of police and non-police reports with the DLNL will help to reduce the burden on police operations and, in turn, enhance police response capabilities.
- KSP Keywords
- Data length, F1-score, Normalized data, South Korea, text data