Name
Abstract | ||
|---|---|---|
Short messaging service (SMS) is one of the fastest-growing telecom value-added services worldwide. However, mobile message spam is a side effect for ordinary mobile phone users that seriously troubles their daily life and, as a result, threatens the revenue of telecom operators. In this paper, we present an SMS antispam system that combines behavior-based social network and temporal (spectral) analysis to detect spammers with both high precision and recall. The system infrastructure and the proposed approximate neighborhood index solution, which solves the scalability issue of social networks, are described in detail. Experimental results demonstrate that our proposed system achieves excellent discrimination between spammers and legitimates, and even with fixed recall at 95%, the online system and offline detection subsystems maintain a precision of about 98% and 99.5%, respectively. |
Year | DOI | Venue |
|---|---|---|
2010 | 10.1147/JRD.2010.2066050 | IBM Journal of Research and Development |
Keywords | Field | DocType |
social network,system infrastructure,ordinary mobile phone user,sms antispam system,online system,proposed system,high precision,fastest-growing telecom value-added service,fixed recall,behavior-based sms antispam system,mobile message spam | Revenue,Social network,Computer security,Computer science,Telecom operators,Precision and recall,Mobile message,Mobile phone,Scalability | Journal |
Volume | Issue | ISSN |
54 | 6 | 0018-8646 |
Citations | PageRank | References |
14 | 0.72 | 9 |
Authors | ||
9 | ||