The anonymity and decentralization of Bitcoin make it a significant medium for illicit transactions, posing challenges for traditional detection methods in handling complex transaction network structures. This study proposes a graph neural network model based on a pre-trained Conditional Variational Autoencoder (CVAE) to enhance the efficiency and accuracy of Bitcoin illicit transaction detection. The model generates K−1 feature vectors through the CVAE, which have the same number as the input features, and then combines these generated K−1 feature vectors with the original feature vector to ultimately form K feature vectors. Each feature vector undergoes multi-channel aggregation and max pooling, resulting in multiple feature vectors. These vectors are subsequently processed through linear layers and layer normalization, followed by another round of max pooling to obtain a global feature vector. Finally, the feature vectors are further processed through graph convolutional layers and linear layers to generate the final classification result. The model integrates input feature vectors at the output layer through a skip mechanism. Experimental results demonstrate that this model performs excellently in Bitcoin illicit transaction detection, significantly improving detection accuracy and robustness.

As digital currencies, notably, Bitcoin, gain widespread adoption, the detection of illicit transactions poses a pressing challenge that requires prompt resolution. This research introduces a novel approach for detecting illicit Bitcoin transactions that integrates diverse features to enhance both detection efficiency and accuracy. Firstly, a comprehensive feature set is constructed by amalgamating conventional data features with those uniquely derived from LSTM, RandomWalk, and PageRank algorithms, enabling the capture of intricate patterns within transaction data. Secondly, to address the class imbalance inherent in Bitcoin transaction datasets, FocalLoss is adopted as the loss function, strengthening the model’s ability to discern minority classes (i.e., illicit transactions). Finally, the model is validated on the Elliptic dataset using a multilayer perceptron (MLP) architecture with a single hidden layer, and its performance is compared with current mainstream Bitcoin illegal transaction detection models (GAT, GCN). Experimental results demonstrate that the proposed method achieves significant improvements in crucial metrics such as F1 score and recall rate compared to traditional methods, validating the effectiveness of the multi-feature fusion strategy and the utilization of FocalLoss in tackling the challenge of illegal Bitcoin transaction detection.

As technology advances, mobile device forensics becomes increasingly challenging in the context of constantly updated operating systems and enhanced data encryption techniques. This paper, taking the TikTok application data extraction from a Huawei mobile phone in an actual case as an example, delves into the difficulty of extracting data from specific applications in new Android phones. To tackle this challenge, this study proposes a method utilizing a root-privileged phone to clone data from the source device and subsequently extract the required information, thereby achieving successful data retrieval. Furthermore, recognizing the inefficiencies and time-consuming nature of traditional manual timestamp conversion methods during targeted database analysis, this study has developed a novel database retrieval tool. This tool automates the process of swiftly retrieving and analyzing data from key time periods across multiple databases within a predefned directory, significantly enhancing processing speed and efficiency. Thus, our study not only offers a solution to the challenges of data extraction and analysis but also serves as a valuable methodological reference for mobile device forensics in similar cases.