Andi Ashari

Tech Voyager & Digital Visionary

Corruption System Development Based on Indonesia’s Corruption Perception Index

File Typeapplication/pdf
File Size298,949 KB
Published At11/15/2017
File Hashaa3faec43e6e8421585b984d79d16fec
Download Link

In my research, "Corruption System Development Based on Indonesia’s Corruption Perception Index," I, Andi Muhammad Muqsith Ashari, aimed to tackle the pervasive issue of corruption in Indonesia by leveraging the capabilities of artificial intelligence. Guided by Lili Ayu Wulandhari, S.Si., M.Sc., Ph.D., a distinguished AI lecturer at BINUS University, we embarked on developing a system that utilizes online news as a reflection of the current corruption state. This project not only highlights the significant role of government in combating corruption but also underscores the innovative application of AI in public governance, aiming to enhance transparency and accountability across Indonesia's regions.


This research focuses on developing a system to map corruption in Indonesia, utilizing online news to reflect the nation's current state. The prevalence of corruption among civil servants necessitates focused governmental attention on every region to prevent corruption. The study employs the Naïve Bayes classifier to categorize news into corruption-related and non-corruption segments. N-Gram and Hash Table methodologies are then used to geographically map these cases across Indonesia’s administrative divisions. The outcomes reveal a 100% accuracy in classification and 85% in geographic mapping.


Indonesia, Southeast Asia's largest economy, faces significant challenges due to corruption among government officers, undermining economic growth and good governance. This research aims to leverage mass media coverage as a metric for Indonesia's Corruption Perception Index, proposing a system to aid the government in analyzing and managing corruption regionally. This study is an extension of previous work on mapping corruption cases based on the Corruption Perception Index.


The methodology encompasses web crawling, web scraping, and cron scheduling to collect data from seven prominent Indonesian news websites. This approach ensures a comprehensive dataset, which is then processed through text mining techniques, including tokenizing, stopping, and stemming, to prepare for Naïve Bayes classification. The classified data are further analyzed to extract geographical information for mapping.

Results and Analysis

The system successfully classified 60 news articles (30 about corruption and 30 not related to corruption) with 100% accuracy. For geographical mapping, the accuracy was 85%, indicating the system's effectiveness in identifying corruption news and their locations. However, the reduced accuracy in mapping suggests areas for improvement in accurately identifying geographical details from the news content.