An Efficient Unsavory Data Detection Method for Internet Big Data

摘要

With the explosion of information technologies, the volume and diversity of the data in the cyberspace are growing rapidly; meanwhile the unsavory data are harming the security of Internet. So how to detect the unsavory data from the Internet big data based on their inner semantic information is of growing importance. In this paper, we propose the i-Tree method, an intelligent semantics-based unsavory data detection method for internet big data. Firsdy, the internet big data are mapped into a high-dimensional feature space, representing as high-dimensional points in the feature space. Secondly, to solve the "curse of dimensionality" problem of the high-dimensional feature space, the principal component analysis (PCA) method is used to reduce the dimensionality of the feature space. Thirdly, in the new generated feature space, we cluster the data objects, transform the data clusters into regular unit hyper-cubes and create one-dimensional index for data objects based on the idea of multi-dimensional index. Finally, we realize the semantics-based data detection for a given unsavory data object according to similarity search algorithm and the experimental results proved our method can achieve much better efficiency.