Blockchain Model of Sensitive Information Identification in Social Online Games

Abstract

This paper studies the sensitive information recognition and processing system in social online games. A real-time data mining framework for sensitive information on online games in the blockchain mode is proposed. A large-scale social online game chat room database is selected for testing, with SKLearn library based on Python is used for data preprocessing. The model training and verification are implemented through tensorflow. The results show that the accuracy of TextCNN is over 5% higher than that of Naive Bayes and CNN models for short text recognition in online game context, so TextCNN model can meet the requirements for high efficiency and accuracy of sensitive information recognition in large-scale online games. The conclusion proves that the blockchain-based online game sensitive information mining framework has the advantage of alliance chain hierarchical management, the characteristics of the distributed ledgers leaving traces in the whole process of information dissemination, and enhanced contract enforcement through the smart contracts and consensus mechanisms.