Abstract
In this study, I've suggested how to improve the function of web-agents to find out the web-document users prefer. Web-agents employ TFIDF, which considers all the worked used in a document as equal in improtance to find out users' preferences. Web-documents like HTML, however, make visual differences by using different sizes of letters and highlighting them based on importance of words. In this study, I've attempted to improve the functions of the web-agents by differentiating the weight of each worked in accordance with the visual importance of each paragraph. To enhance functions, I've suggested how to make a profile from each paragraph to be consolidated later. As to suggested algorithms, I've tested their effects by comparing the established TFIDF algorithm with the function which helps users find documents they prefer.
