A Multi-agent System Based on Evolutionary Learning for the Usability Analysis of Websites.

In this chapter we propose a novel multi-agent system for the semi-automated study of the usability of websites, an increasingly critical issue given the ubiquity of the Web and its technological evolution. The proposed system constructs a key phrase-based model of the users trying to reach one URL from another, simulates the browsing process, and analyses the web pages in the path. The resulting usability analysis is focused on issues such as navigation paths, links, page content, HTML coding, and accessibility. Our system automatically draws usability conclusions and suggestions and also presents significant data in support of the human usability expert. The architecture of the system consists of rule-based reactive agents subject to evolutionary processes. The application of evolution allows the agents to explore possible solutions in a more realistic way than either exhaustive or arbitrary examinations. [ABSTRACT FROM AUTHOR]