Development of a Web Based Corruption Case Mapping Using Machine Learning with Artificial Neural Network

Cumulative state loss over the years caused by corruption in Indonesia has reached a fantastic number of 15 Billion usn up until 2016 [1]. To imagine the severances of it, 10.000 KM of highway can be built with that much amount of fund. Several responses has been done by Indonesian government to fight corruption from both prevention and persecution. This work focuses on the development of a web application aimed to provide insight to corruption case per province in Indonesia. The web application was developed using Machine Learning, specifically Backpropagation Artificial Neural Network (ANN). Web crawling and web scraping techniques are used to gather news content from 7 major news portal in Indonesia. Accuracy is measured by comparing correct prediction by ANN to its true value. Upon finding corruption news, data is saved and further analysis is done to establish the number of corruption news per region. Finally, the output is visualized using Google Map API. The purpose of this work is to provide regional depiction of corruption case to give further insight to be used by decision makers. Upon using this web application, objectivity of program development is expected to increase. The final output of the web application is a map of corruption case in Indonesia per region. The accuracy of ANN used in this work to classify corruption and non-corruption news is 96.91 % using Sigmoid Activation Function. © 2018 IEEE.