Detecting Phishing Scams on Ethereum based on Transaction Records

ABSTARCT :

With the increasing popularity of blockchain technology, it has also become a hotbed of various cybercrimes. As a traditional way of scam, the phishing scam has new means of scam in the blockchain scenario and swindles a lot of money from users. In order to create a safe environment for investors, an ecient method for phishing detection is urgently needed. In this paper, we propose a three steps framework to detect phishing scams on Ethereum by mining Ethereum transaction records. First, we obtain the labeled phishing accounts and corresponding transaction records from two authorized websites. According to the collected transaction records we build an Ethereum transaction network. Then, a network embedding method node2vec which can extract the latent features of accounts is used for subsequent phishing classification. Finally, to distinguish whether the account is a phishing account, we adopt the one–class support vector machine (SVM) to classify. The experimental result demonstrates that F-score of our phishing detection method can achieve 0.846, which verifies the validity of our model. To the best of our knowledge, this is the first work that investigates the phishing scams on Ethereum based on transaction records.

EXISTING SYSTEM :

? An existing data set was used that contained car trajectories and vehicle telemetry (such as speed, acceleration, heading, etc.). ? However, with the phishing scam revealed, in-transactions become rare, or even nonexistent. ? This leads to in-transactions are concentrated in a small period for a phishing address, and the feature can grasp this characteristic very well. ? A systematic survey of the existing Distributed Denial of Service (DDoS) attacks detection and mitigation strategies in Software-Defined Networking (SDN).

DISADVANTAGE :

? Detecting phishing scams in the blockchain ecosystem is a real and critical problem. ? Different from traditional currencies which are usually issued by authorized financial institutions, the issue and operation of cryptocurrencies are not under the regulation of any authorized organizations. ? SVM and DT are considered efficient in many classification problems of class imbalance. ? To help dealing with this issue, in this study, we propose a systematic approach to detect phishing scams in the Ethereum ecosystem.

PROPOSED SYSTEM :

• We propose a graph-based cascade feature extraction method, which can conveniently extract rich transaction structure information and form a feature set with a good classification effect. • We propose a systematic approach to detect phishing scams in the blockchain ecosystem, and take Ethereum as an example to verify the effectiveness. • We propose a new model integration algorithm, namely the Dual-sampling Ensemble algorithm, which can be used for classification problems with a high level of class imbalance. • We propose a TG-based cascade feature extraction method for phishing account identification.

ADVANTAGE :

? Time plus amount features achieve better performance than only used time or amount features. ? Moreover, time features only method presents better performance than amount features only method, which indicates that the importance of time features. ? The performance of their model was evaluated presenting the accuracy, precision, recall and F1-score for each data set. ? The performance of their model was presented by calculating accuracy for different numbers of participants, and the percentage of modified labels that indicate the strength of the poisoning attack.