Page 32 - bookofabstract_vj5_final(1)

Recommender System Based on Empirical Study of Geolocated Clustering and
Prediction Services for Botnets Cyber-Intelligence in Malaysia

Nazri Ahmad Zamani and Aswami Ariffin

Abstract

A recommender system is becoming a popular platform that predicts the ratings or
preferences in studying human behaviors and habits. The system is widely used
especially in marketing, retailing and product development. The system responds to
users preferences in goods and services and gives recommendations based on
Machine Learning algorithms deployed catered specifically for such services. The same
recommender system can be built for predicting botnets attack. Via our Integrated
Cyber-Evidence (or ICE) Big Data system, we build a recommender system based on
collected data on telemetric Botnets networks traffics. The recommender system is
trained periodically on cyber-threats enriched data from Coordinated Malware
Eradication & Remedial Platform system (or CMERP), specifically the geolocation and
the timestamp of the attacks. The machine learning is based on K-Means and DBSCAN
clustering. The results is a recommendation of potential attacks from a given
geolocation coordinates.

28