Construction and Evaluation System of Machine Learning Language Model for Digital Network Resources
Jiayi Liu, Lingyue Meng · 2025
In view of the inefficiency of traditional digital network resource management and search, this paper introduces a machine learning language model, which aims to improve the efficiency and accuracy of resource retrieval through intelligent resource indexing, classification and recommendation system. By designing precise models and real-time evaluation methods, it ensures that resources can be processed and allocated quickly and accurately in large-scale network data. First, in the system design stage, a neural network-based framework was constructed, and BERT (Bidirectional Encoder Representation Transformer) was used to train a large number of digital network resources. By extracting resource feature information, a classification model suitable for network resources was constructed. Then, a CNN-LSTM (Convolutional Neural Network-Long Short-Term Memory Network) combination was used to achieve accurate semantic understanding and automatic resource classification during feature extraction and processing. Finally, in order to further optimize the system performance, a deep Q network (DQN) was introduced to continuously improve the recommendation accuracy and response speed through an adaptive optimization process. This reinforcement learning model can adjust the strategy in real time according to the user's interaction data to ensure the efficient operation of the system in a complex network environment. The CNN-LSTM fusion model performed particularly well in the resource classification task, with an overall accuracy rate of over 90%, especially in tasks 2, 5, and 10, where the accuracy rates climbed to 92.5%, 93.2%, and 93.0%, respectively. By introducing a combination of deep learning and reinforcement learning, the system provides a solution to the efficiency issues in traditional digital network resource management, significantly improving the efficiency of resource retrieval, classification, and recommendation.