Analysis and Implementation of Ontology Based Text Classification on Criminality Digital News
Fauzia Rahma, Dwitika Diah Pangestuti, Anisa Herdiani, Nungki Selviandro · IOP Conference Series Materials Science and Engineering · 2019
Abstract This research, we search the trend by utillize all information about the criminality type. When the trend has found, the possibility of crime in West Java will decrease. We need to analyse the data to get the information about the criminality trend by using Ontology based Text Classification Method. A news will be grouped into some criminality classes based on the relation around that groups so the process of classification can be done simply and specifically. The criminality trend of West Java is Property Crime with 47,5% of occurrences. We used several testing method such as F1 Score, Precision, Recall, and Accuracy. Based on the methods, the performance of system is running well or not in different point of view. The researcher obtained satisfactory result with F1 score 87,05%, accuracy 86,74%, recall 100% and precision 77,08% with using comparison of composition crime dataset and non-crime dataset is 40 : 40. It happens because testing needs to be done by considering the number of both datasets. The more balance the comparison while do preprocess, the higher accuracy that will get. In order to know whether the system is able to clarify the information accurately and can separate the non-criminal news dataset.