Andi Ashari

Tech Voyager & Digital Visionary

Development of Interactive Mapping Systems in Cases of Criminality in Indonesia Based on Online News Perception Using Artificial Neural Network (ANN) Method

File Typeapplication/pdf
File Size2,213,300 KB
Published At3/19/2018
File Hashc7143bdfd20094de34b1d38c66b5924c
Download Link

As Andi Muhammad Muqsith Ashari, a student at BINUS University, School of Computer Science, specializing in Computer Science - Artificial Intelligence, my thesis project aimed at harnessing the potential of ANN in analyzing and mapping crime through online news perception. Under the mentorship of Lili Ayu Wulandhari, S.Si., M.Sc., Ph.D., I ventured into developing a system that could not only classify news into predefined crime categories but also visually represent the spread and intensity of crime across Indonesia through a heat map. This endeavor was not just an academic pursuit but also a practical solution aimed at enhancing public safety and awareness.


Despite ongoing efforts, the number of criminal cases in Indonesia remains unchanged annually, with mass media continuously highlighting incidents across the country. This situation calls for more focused attention from the government and relevant bodies to each region to mitigate crime rates. An effective strategy involves mapping criminal activities based on online media reports, considered a mirror of Indonesia’s current state. This thesis introduces an interactive mapping system developed using Artificial Neural Network (ANN) to classify online news articles into specific crime categories. The process begins with text processing to standardize data input, utilizing N-Gram and Hash Table techniques for accurate provincial mapping in Indonesia. The findings reveal a perfect accuracy rate (100%) in both the ANN’s classification phase and the mapping stage, highlighting the system’s effectiveness in crime perception analysis.


The methodology section outlines the comprehensive approach taken to process text data, prepare it for ANN analysis, and ensure the accurate classification of crime news. The utilization of N-Gram and Hash Table techniques was critical in mapping crime cases based on Indonesia’s provincial levels, demonstrating the system’s capability to handle and analyze complex data sets effectively.

Findings and Conclusion

The ANN classification method, employing the Backpropagation Neural Network algorithm, achieved an impressive 92.80% accuracy in classifying news into three predetermined classes after text processing.

The optimal use of parameters in the BPNN algorithm led to an average accuracy rate of 96.45%, highlighting the efficiency of using a small number of hidden neurons and a high learning rate, despite the large number of iterations required.

The location identification using N-Gram and Hash Table methods successfully mapped news articles to their respective provincial locations in Indonesia with 100% accuracy, although it required significant computational resources.


Suggestions for future research include diversifying training and testing documents to improve accuracy, adding filters to avoid duplicate news from different sites, and optimizing N-Gram sizes to reduce memory usage in the news identification process. This thesis contributes to the field of AI in public safety, offering a novel approach to understanding and visualizing crime through data analysis and interactive mapping.