Enhancing internet experience of visually impaired persons by means of dynamic highlighting and selective reading

This paper proposes a research-based tool to assist visually impaired persons (VIPs) in using the Internet via screen readers. The proposed tool is inspired by research on modeling web use and model-based highlighting. This tool assists VIPs in selecting goal-relevant information on web pages. A computational cognitive model simulates the VIPs’ Internet use. An intelligent agent capable of dynamic highlighting and selective reading based on efficient machine learning algorithms runs alongside the (simulated) user. The agent learns from interacting with the cognitive model and the information space. This agent is implemented in an adaptive interface that takes, expands and updates a user goal, finds goal-relevant information and suggests it to the (simulated) user in an appropriate way. The proposed tool could be applied in situations that require handling information overload with limited perceptive and cognitive capabilities.