Abstract: Network firewall is the connection between the computer and the network channel, in the filter network virus, to protect the security of computer users there is an important role. The performance of the internal data filtering protection algorithm of the network firewall directly affects the ability of filtering the virus. However, most of the network firewall data are based on the pre-set security policy. The data packets inside and outside the network are checked and checked. There is a possibility of misjudgment.In this paper, an internal data filtering and protection algorithm based on community trust is proposed. Firstly, the internal database of network firewall is modeled as a hierarchical tree, and the problem of dangerous data acquisition in network firewall is transformed into tree traversal problem. By comparing the attribute of the tree, the query space is reduced. The accuracy and efficiency of the extraction of the dangerous data within the network firewall are improved by calculating the correlation of the attribute values. Then, the data trust of the network firewall is determined by calculating the data characteristics. The trust of the target data evaluation, and then to achieve the network firewall internal data filtering protection. The simulation results show that the proposed method can improve the accuracy of network firewall filtering protection effectively, improve the efficiency of filtering and protecting the data within the network firewall, and has strong practicability.