Multi-dimensional Intelligent Network Asset Identification
Yi Xiong, Xiaoyan Zhang, Ling Pang, Zhechao Lin · 2021
In the field of cyberspace detection and identification, how to improve the accuracy of asset identification is always a great challenge for people. This paper proposes a multi-input fusion model based on TextRCNN, which uses multi-dimensional features to identify asset types, so as to improve the accuracy of the identification. Firstly, we analyze many kinds of assets and select the protocols with symbolic characteristics. The data set is obtained based on the detection results of the selected protocols. Then, the deep learning multi-input fusion model is constructed for training. Finally, on the basis of active detection and identification technology, this model is used to improve the accuracy and credibility of asset type identification. The characteristics of this method are as follows: the fusion network can process multi-protocol input data at the same time, so as to build a multi-dimensional identification model; the identification of key attributes of assets is based on multi-dimensional characteristics, and no longer depends on the identification result of a single port. The experimental results show that the accuracy of the multi-dimensional intelligent recognition method can reach 89.48%, which can effectively identify the asset types.