PaSent - Electronic Health Advisor

01.03.2004

Abstract
Made publicly available in the early nineties, the World Wide Web was an immediate success. Accessibility increased and everybody wanted to be online. Reaching the mid nineties the interest went from being large to enormous. It was a digital revolution and distribution of information had never been easier. Many different communities have settled on the Web, health being one of the largest.

Today health- and medical literature is widely available on the Internet. As the number of pages within this community grows, finding relevant documents get more and more difficult. The need of search tools is obvious and has become a multi-million dollar industry. With the enormous amount of information available the relevance and correctness becomes essential. Evaluating search result manually is time consuming and difficult. This has changed the focus of the search industry from locating information to filtering and categorizing information.

One system trying to include the filtering process in a search mechanism is PaSent. PaSent is a search and filtering system specialized for health and medical information. It uses a medical document and a context model as input and generates a neural network. This neural network is used to filter and rate documents located on the Web.

This thesis describes our work on PaSent. We have designed and implemented a user interface. Later we conducted a usability test to reveal possible usability flaws. In addition we have redesigned the overall architecture of the previous version and made adjustment suiting our implementation.