Improving Website Usability with Latent Semantic Analysis
Sarah A. Nuehring, Peter W. Foltz · Proceedings of the Human Factors and Ergonomics Society Annual Meeting · 2006
When people search for information on the Web, they rely on link descriptions to decide where to look. If the descriptions are not sufficiently accurate, people will have a hard time navigating. This difficulty is what is known as the vocabulary problem. The overall goal of this study was to develop a way of making the vocabulary problem less probable on websites: first by improving the information scent of links in websites, and second by investigating the relationship between information scent and website navigability. Information scent was defined as the semantic similarity between a link description and its target information, quantified using Latent Semantic Analysis (LSA). Lists of keywords related to web pages were obtained from human participants and from LSA, and then the keywords were used to make modifications to the links on web pages. Participants showed faster navigation when using the modified websites compared to using the original websites.