Research On Network Information Resource Evaluation Model Based On Neural Network

Network Information Resources Evaluation is one of the focuses on the current network information resources management research. In order to provide more effective services for users to using network information resources, this paper tries to use the neural network model to design a set of network information resource evaluation system in the qualitative and quantitative aspects.According to network information resources evaluation principles and standards, this paper analyzed the network information resources evaluation status at the present time, and proposed the combination of qualitative and quantitative evaluation index system, and designed 16 evaluation indexes. Then, this paper applied the neural network technology to the network information resource evaluation index system, analyzed the structure of BPNN(back propagation neural network), BP algorithm method, flow, etc. At the same time, there are a lot of disadvantages and deficiencies in the traditional BP algorithm:such as slow convergence speed, easy to fall into local minimum point, the number of hidden layers and nodes and other issues, this paper proposed solutions to these problems. According to testing and comparing several improved methods about BP algorithm, this paper decided to go with a single hidden layer BP network model, and used the Levenberg-Marquardt algorithm to optimize the BP network. Finally, we used the MATLAB software to simulate the established network of information resources evaluation based on BPNN model. Contrasting to traditional statistical methods in network information resources evaluation, simulation results showed that network information resources evaluation based on BPNN model had certain advantages in recognition rate of evaluation classification, system time-consuming and fitting precision, etc, and showed the effectiveness of the model.