Generation of Realistic Navigation Paths for Web Site Testing Using Recurrent Neural Networks and Generative Adversarial Neural Networks

摘要

A robust technique for generating web navigation logs could be fundamental for applications not yet released, since developers could evaluate their applications as if they were used by real clients. This could allow to test and improve the applications faster and with lower costs, especially with respect to the usability and interaction aspects. In this paper we propose the application of deep learning techniques, like recurrent neural networks (RNN) and generative adversarial neural networks (GAN), aimed at generating high-quality weblogs, which can be used for automated testing and improvement of Web sites even before their release.