Metabrain: Web Information Extraction and Visualization

Nowadays, a hitherto unseen amount of information can be found in the World Wide Web. While available, this information is fragmented among different web sites. This is especially true for implicit knowledge, not directly written in any one site, but arising from patterns and interactions between pages. For instance, the number of search results for a particular query string might be a meaningful indicator of its popularity or overall interest. Our research focuses on the design of an interface that allows end-users to access implicit information. A prototype application, Metabrain, embodies our solutions and makes it possible to mine the web for statistically relevant patterns, with the help of simple and straightforward algorithms and user interface. To help the users make sense of that information, Metabrain then allows custom visualizations to be crafted. User studies show that users can search for relevant information up to four times faster than using traditional Web search engines alone. A system usability scale questionnaire confirms the interface is usable and effective.