Abstractly, the information to be displayed consists of a large number of objects, a set of features that are likely to be of interest to the user, and some function that measures the relevance level of every object to the various features. The goal is to provide the user with a concise and comprehensible visualization of that information.

For the type of applications the talk concentrated on, the exact relevance measures of the objects are not significant. This enables accuracy to be traded for a clearer display. The idea is to 'flatten' the multidimensionality of the feature space into a 2D 'relevance map', capturing the inter-relations among the features, without causing too many ambiguous interpretations of the results. To better reflect the nature of the data and to resolve the ambiguity, the given set of features is refined, and the notion of composed features is introduced.

The prototype, built to visualize information extracted from the World-Wide Web, loads information from query engines and automatically categorizes and clusters the information, allowing the user to visualize the information in a hyperlinked information 2D map. This provides a navigational aid for the 'lost in hyperspace' syndrome.

Joint work with Daniel Cohen-Or and Tova Milo, presented at the IEEE Visualization 1997 conference.

The speaker described a method that can produce reflection images (and, with minor modifications, refraction images) at interactive rates. These images can be combined with a primary image containing ordinary view-independent lighting (local shading and textures) to create images having the distinctive 'ray tracing' look. Because all types of images can be generated by standard graphics hardware, the result is unprecedented realism in real time.

Joint work with Ari Rappoport, part of which is to be presented at the SIGGRAPH 1998 conference in Orlando.

Joint work with Gershon Elber, to be presented at the 8th IMA Conference on Mathematics of Surfaces, Birmingham, September 1998.

With a system of parallel coordinates, a one-to-one mapping between subsets of N-space and subsets of 2-space is obtained. Lines in N-space are represented by N-1 indexed points. In fact, all p-flats (planes of dimension p in N-space) are represented by indexed points, where the number of indices is one more than the object's dimension. The representations are generalized to enable the visualization of polytopes and certain kinds of hypersurfaces, as well as convex objects in R^N. Synthetic construction algorithms involving intersections, proximity, an interior point algorithm and 'line topologies' of interest in computer vision were presented. The talk also included interactive demonstrations of applications to process control, visual data mining (improving yields in VLSI production, finance, retailing etc.), collision avoidance algorithms for air traffic control, optimization and nonlinear modeling.

The speaker presented an algorithm for determining a conservative superset of the primitives visible from an epsilon-neighborhood of a given viewpoint. Having such an epsilon-superset, the client is free to render the model independently of the server, as long as he or she moves within that epsilon-neighborhood.

Joint work with Daniel Cohen-Or, to be presented at the Graphics Interface 1998 conference.