Augmenting Telephone Spam Blacklists by Mining Large CDR Datasets

Liu, Jienan; Rahbarinia, Babak; Perdisci, Roberto; Du, Hong; Su, Li

doi:10.1145/3196494.3196553

Cited by 8 publications

(12 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The statistics-based detection systems monitor different call statistics of the subscriber during and after the call. These approaches first collect statistics from the raw call detailed records or signaling messages and then apply data mining to characterize the behavior of the caller [128,129,130].…”

Section: Robo or Telemarketing Call Detection Systemsmentioning

confidence: 99%

Consumer-facing technology fraud: Economics, attack methods and potential solutions

Ali

Azad

Centeno

et al. 2019

Future Generation Computer Systems

View full text Add to dashboard Cite

show abstract

Section: Robo or Telemarketing Call Detection Systemsmentioning

confidence: 99%

Consumer-facing technology fraud: Economics, attack methods and potential solutions

Ali

Azad

Centeno

et al. 2019

Future Generation Computer Systems

View full text Add to dashboard Cite

show abstract

“…Therefore, given a large and distributed user population, it is reasonable to consider heavy hitters as candidate spammers. In other words, a caller ID that is reported as unknown by a significant fraction of participating smartphones satisfies the volume and diversity features used in previous work [22,23], and can be considered for blacklisting.…”

Section: Problem Definition and Approachmentioning

confidence: 91%

“…Protecting the IP address and identity of users may be achieved via other security mechanisms that are outside the scope of this work. Approach Overview According to recent work on phone blacklisting [22,23], it is clear that most spammers will tend to call a large number of users, in an attempt to identify a subset of them who may fall for a scam. Therefore, given a large and distributed user population, it is reasonable to consider heavy hitters as candidate spammers.…”

Section: Problem Definition and Approachmentioning

confidence: 99%

Towards a Practical Differentially Private Collaborative Phone Blacklisting System

Ucci

Perdisci

Lee

et al. 2020

Annual Computer Security Applications Conference

Self Cite

View full text Add to dashboard Cite

“…Azad et al [17] utilize the K-mean clustering algorithm to mark the caller as a spammer or a non-spammer. Liu et al [38] discovered the telephone numbers involved in spam campaigns by using the unsupervised and supervised machine learning methods along with the known spam phone numbers to find out new spammers. Sharbani et al [43] estimate the effectiveness of spam blacklists by measuring their ability to block future unwanted phone calls.…”

Section: Related Work and Motivationmentioning

confidence: 99%

Socioscope: I know who you are, a robo, human caller or service number

Azad

Alazab

Riaz³

et al. 2020

Future Generation Computer Systems

View full text Add to dashboard Cite

Telephony technologies (mobile, VoIP, and fixed) have potentially improved the way we communicate in our daily life and have been widely adopted for business and personal communications. At the same time, scammers, criminals, and fraudsters have also find the telephony network an attractive and affordable medium to target end-users with the advertisement, marketing of legal and illegal products, and bombard them with the huge volume of unwanted calls. These calls would not only trick call recipients into disclosing their private information such as credit card numbers, PIN code which can be used for financial fraud but also causes a lot of displeasure because of continuous ringing. The fraudsters, political campaigners can also use telephony systems to spread malicious information (hate political or religious messages) in real-time through audio or text messages, which have serious political and social consequences if malicious callers are not mitigated in a quick time. In this context, the identification of malicious callers would not only minimize telephony fraud but would also bring peace to the lives of individuals. One way to classifies users as a spammer or legitimate is to get feedback from the call recipients about their recent interactions with the caller, but these systems not only bring inconvenience to callees but also require changes in the system design. The call detail records extensively log the activities of users and can be used to categorize them as the spammer and non-spammer. In this paper, we utilize the information from the call detailed records and proposed a spam detection framework for the telephone network that identifies malicious callers by utilizing the social behavioral features of users within the network. To this extent, we first model the behavior of the users as the directed social graph and then analyze different features of the social graph i.e. the Relationship Network and Call patterns of users towards their peers. We then used these features along with the decision tree to classify callers into three classes i.e. human, spammer and call center. We analyzed the call record data-set consisting of more than 2 million users. We have conducted a detailed evaluation of our framework which demonstrates its effectiveness by achieving acceptable detection accuracy and extremely low false-positive rate. The performance results show that the spammers and call center numbers not only have a large number of non-repetitive calls but also have a large number of short duration calls. Similarly, on the other hand, the legitimate callers have a good number of repetitive calls and most of them interacted for a relatively long duration.

show abstract

Augmenting Telephone Spam Blacklists by Mining Large CDR Datasets

Cited by 8 publications

References 10 publications

Consumer-facing technology fraud: Economics, attack methods and potential solutions

Consumer-facing technology fraud: Economics, attack methods and potential solutions

Towards a Practical Differentially Private Collaborative Phone Blacklisting System

Socioscope: I know who you are, a robo, human caller or service number

Contact Info

Product

Resources

About