Classify and predict web user behaviour using butterfly optimization and recurrent neural network.

Classifying user browsing behavior is an essential task to put suitable information on the web. Also, the browsing behavior is stored in the web server logs, which are used for identifying and classifying commonly assessed patterns of web users. This work designs a Butterfly-based Recurrent Neural Scheme (BbRNS) for accurate classification of the web user browsing behavior based on the URLs. It involves preprocessing, feature extraction, and classification. Generally, preprocessing removes the error, and feature extraction is employed to extract the web user browsing URLs. Then, updating the fitness function in the classification layer for accurate prediction of the browsing behavior of web users also enhances the performance of user behavior detection. Consequently, the developed framework is implemented using a Python tool, and the parameters of the current research work are evaluated with prevailing assignments. The designed model gained 98.8 accuracies, 97.5% recall, and 98% precision for predicting and classifying web user behavior. The experimental result shows improved accuracy for classifying web user browsing behavior with less error rate. [ABSTRACT FROM AUTHOR]