Currently RIA technologies are embodied in a testbed, called RealHunter™, a real-estate application that helps users find residential properties.
Here are several recorded demos that you can download and play with Microsoft Media Player version 9.0 or higher. To get the best effect, you may want to play it on a display with resolution 1280x1024 or higher. Turn on your speaker to hear speech input or output.
- Overview demo (download AVI file here, 9.5 megabytes)
This demo provides the overview of RIA technologies applied to a residential real estate application, which helps potential buyers to look for real estate properties. - Overview demo with speech input (download WMV file here, 2.5 megabytes)
Our system can also run with a speech recognizer that allows a user to express his/her information requests using speech and gesture. - Adaptation demo (download WMV file here, 3.2 megabytes)
To allow users and systems to adapt to each other's expressions over time, this demo shows how a system teaches the user to modify his/her request (e.g., using concrete constraint such as city population to define fuzzy expression like "small city") and how the system gradually learn the user expressions over time (e.g., understanding "good school districts"). - Visual context management demo (download AVI file here, 21.8 megabytes)
Since RIA and users engage in a continuous conversation, RIA must dynamically maintain the proper visual context to allow users to easily integrate all relevant information across interaction turns. This demo shows how RIA dynamically manages the visual display context in various situations (e.g., maintaining visual and semantic continuity).
We extend RIA technologies to develop NIMBLE, an intelligent, mixed-initiative interactive system supporting information investigation. Unlike existing visualization systems, which support isolated visual investigation in discrete steps, NIMBLE supports a coherent and nonlinear visual investigation by dynamically maintaining a visual analytic context. It visually represents evolving analytic threads that are visually linked to the corresponding investigation results. Using the context, analysts can dynamically link information discovered at different steps, and evaluate information in different analytic contexts.
- Demo (download AVI file here, 40 megabytes)
In this video, We use a set of examples to illustrate how a user works with NIMBLE to dynamically build her analytic context as an investigation develops. We also show how HARVEST uses such context to help the user find relevant information. Our main scenario is on the investigation of an alleged political fraud. We are using the IEEE VAST 2006 contest data, including phone records, voter registry, and local news reports.