The Feature has an an article by Mark Frauenfelder:
The most advanced mobile-based Semantic Web project is taking place at Carnegie Mellon University. Called MyCampus, the project is headed up by professor Norman Sadeh, director of the Mobile Commerce Lab at Carnegie Mellon University’s School of Computer Science in Pittsburgh. MyCampus was specifically set up to develop context-aware mobile services for the university’s community. The system runs on PDAs and across 700 WLAN access points located around the university, and it is used daily by 3,000 people to help them study, socialize, plan meals, attend events, shop, and engage in extracurricular activities.
MyCampus consists of several task-specific agents that automatically capture contextual information. Each MyCampus user has a database, called a “Semantic eWallet,” which is a repository for users’ personal information, such as class schedules, list of friends and classmates, and lifestyle and event preferences. Location data is generated using Pango’s WiFi access-point triangulation. All the data is marked with Semantic metadata so that MyCampus agents can make use of it. User’s can set access privileges to allow certain people to know where they are at any given moment, or what their schedule for the upcoming week is.
One of the most popular applications is the restaurant concierge agent, which recommends places based on a user’s dining tastes, schedule, location, and weather conditions. If there’s a storm brewing, the concierge will recommend a place that doesn’t require stepping outside, and if the user has a study group meeting in 30 minutes, it’ll suggest a fast food joint within a block or two.
Recently, a group of students at CMU developed an application for MyCampus called InfoBridge, which lets users post and read “virtual posters” about upcoming events. For example, say a user has indicated that she likes track and field events. Shell be notified about events as soon as another person makes a virtual poster about it, unless shes sitting in class. If thats the case, she wont be notified until class is over. If she wants to attend the event, she clicks on a link and, because the data has been tagged with Semantic Web metadata, itll be added it to her calendar. If theres a scheduling conflict, itll notify her and present her with options. All this data exchange is done with agents — no human screen scraping.