Risk Concentration for Context Assessment (RiCCA) of SMS Messages using Danger Theory
Unified with the booming of mobile and Internet technology, spam is one of the serious issues that have been emerged tremendously. The never ending records of loss that caused by spam have initiated and motivated this study. With the purpose to produce a complementary solution for the current safegu...
Saved in:
| Summary: | Unified with the booming of mobile and Internet technology, spam is one of the serious issues that have been emerged tremendously. The never ending records of loss that caused by spam have initiated and motivated this study. With the purpose to produce a complementary solution for the current safeguards, this study aims to assist users in identifying potential harmful messages. The focus of the study is deploying an English text spam in Short Messages Services (SMS) format that usually consists of fraud intention in disguise. Due to the convincing but deceptive contents, users easily get enticed and their lack of awareness about implicit spam's impact loss is one of the most vulnerable factors that lead to numerous cases of fraud, scam and identity theft. This study applied one of the prominent theories from Artificial Immune Systems (AIS), Danger Theory. The behaviour of dendritic cells to sniff danger that caused by a harmful substance in the human body has become the fundamental idea of this theory. Danger Theory is employed in deciphering the risky content of span1 message via mapping its' biological properties in spam environment. Consolidation of this theory with other procedures such as risk assessment and text mining has produced a model namely as Risk Concentration for Context Assessment or RiCCA. This RiCCA prototype is developed from Danger Theory algorithms that is Dendritic Cell Algorithm (DCA) and Deterministic Dendritic Cell Algorithm (dDCA). Through a series of simulation with the deployment of merged dataset from UCI Machine Learning and self-collected spam messages, has indicated RiCCA as a reliable medium for measuring the risk concentration. Both algorithms demonstrated that Danger Theory is feasible in measuring risk of SMS content with more than 90% of accuracy rate, which dDCA outperformed DCA with distinctive result of 94.16% accuracy rate. For future work, the prototype can be further enhanced as one of the mobile application. Moreover, this study can be further applied for a larger size of message context (instead of SMS that is limited to 160 characters) and also tested in other languages. |
|---|