Rule-Based NER for Crime Information Extraction through Online News Site

One of the targets of SDG 16 is crime prevention. To achieve this target, knowledge of crime statistics is essential. In Indonesia, the National Police’s Crime Data Center (Pusiknas Polri) and BPS Statistics Indonesia (Badan Pusat Statistik) released an annual publication containing crime statistics to assess the security conditions. However, these annual publications have limitations in presenting up-to-date statistical data. Therefore, this research explores the potential of online news websites as a source of information to complement the current picture of actual crime information in Indonesia. This study uses a rule-based NER model to extract information from online news content. The model automatically identifies crime-related information within the news articles. The extracted information is then analyzed to obtain an overview of crime-related information in Indonesia. This research successfully identified $\mathbf{1 8, 6 6 5}$ crime news from 53,336 crime news published in 2021-2022 on detik.com. The results of the performance evaluation of the proposed NER model received an average F-measure score of 90.71%. This score shows that the proposed model effectively identifies crime news information from online news websites. Crime information successfully obtained from online crime news includes the distribution of total crime based on crime type and location and the characteristics of age groups involved in crime cases.