Click here to bypass page layout and jump directly to story.=

UC Berkeley >

University of California

News - Media Relations







  Press Releases

  Image Downloads



San Jose tech expo to feature UC Berkeley's latest research
08 March 2001

By Catherine Zandonella


CONVERSATION MAP: Navigating Very Large Scale Conversations

The Conversation Map brings order to the chaos found on newsgroups, the bulletin boards of the Internet. Just as an Internet browser helps users navigate the Web, the Conversation Map is a browser that categorizes and maps Very Large Scale Conversations (VLSCs) such as are found in Usenet newsgroup postings. In the process, said Warren Sack, an assistant professor in the UC Berkeley School of Information Management & Systems, the Map can tell us something fundamental about the nature of communication.

To sort out the massive number of newsgroup postings, the Map scans the content of messages, identifying key words and relationships between posted messages. The browser then generates a screen containing four frames, one that monitors the social network of senders, one that uncovers key themes, one (called semantics) that sorts out words used as metaphors, and a final frame that displays threads, or groups of messages, that are responses to previous postings. By clicking on a node within the social network, a user can see who's posting, who they've replied to, what threads they are following, as well as key themes and metaphors in the conversation. To compute all this, the Map system draws on techniques from computational linguistics and quantitative sociology.

While the tool may someday be used to improve newsgroup search engines like the one formerly operated by (now owned by Google Inc.), the killer ap for now is research on the social and semantic relationships that emerge from Internet discussions. A college professor can use the Map to monitor who in her class is participating in a mandatory online discussion group. A linguistics research can observe how metaphors evolve, while a sociologist can observe the odd behaviors - like flaming, for example - that emerge when thousands of people join a conversation.

DENIM and OUTPOST: Drawing the Design Process into the Digital Age

Web design tools are becoming ever more sophisticated, but despite the high-tech software, designers still start out with a pen and paper in hand. A group of UC researchers has designed a set of tools to bring digital power to the early stages of the design process.

James Landay, assistant professor of computer science, and his students have designed two software programs to move the brainstorming process from the white board to the computer. Their efforts are part of UC Berkeley Group for User Interface Research's focus on freeform interactions with technology rather than more constraining interactions with keyboards and mice.

Two technologies will be on display:

* DENIM: With the wave of an electronic pen, Web designers can turn ideas into prototype pages using DENIM, a software program that makes it easy to sketch the main structure of the site and add links, titles and text. While DENIM works best with a digital tablet and stylus, it also works on a desktop PC with a mouse. DENIM is available for free download at

* The Designer's Outpost: For large Web sites, nothing beats a white board for keeping pages and thoughts organized. The Designer's Outpost extends this process, using an electronic white board and two cameras to track the group design process. As the designers write on the board, their drawings and text - even Post-It notes - are recorded by a camera placed within the board and a ceiling-mounted camera trained on the front of the board.

SMART DUST: Tiny Sensors

Cheap and tiny microelectromechanical sensors known as "smart dust" are under development at the UC Berkeley laboratory of Kris Pister, associate professor of electrical engineering and computer sciences. Smart dust motes contain a complete sensor/actuator/communication system in a cubic millimeter. These tiny devices can be spread throughout a building to monitor energy use and occupancy, within bridges and buildings to monitor structural integrity, or along freeways to follow traffic flow.

Recent innovations include the addition of a foot-long radio-controlled plane can drop the motes from above, scattering them over a wide area. Pister and his students have developed maple seed-shaped pieces of silicon that they can use to gently waft the motes towards the ground.

On display will be a smart dust prototype - a one-cubic inch sensor that utilizes an integrated device for measuring temperature, humidity, pressure, light, tilt, vibration and magnetic fields with a bi-directional radio link, microprocessor and battery. The motes communicate via a tiny operating system (Tiny OS) designed by UC Berkeley computer science professor David Culler and his students.

Related technologies on display include:

* Microflying Insects - Microelectromechanical insects whose wings flap using built-in actuators that simulate insect wing motion

* The Acceleration Glove - A glove that senses motion and could someday replace your keyboard.

More information can be found at:

TELEGRAPH: Beyond Mining the Deep Web: Analysis of Data-Rich Sources

Today's Internet search engines barely graze the surface of the information available on the Internet. Taking Deep Web search engines one further, UC Berkeley associate professors Joseph Hellerstein and Michael Franklin have designed Telegraph, a data retrieval system designed harness streams of live data coming out of the Internet as well as from networked sensors, software, and smart devices. Designed by UC Berkeley

Two examples of Telegraph will be on display:

* Federated Facts and Figures: Want to find out which industries gave the most money to President George W. Bush's 2000 campaign? Or whether home prices in your area are a predictor of which candidate your neighbors voted for? Check the Federated Facts and Figures Web site. Using Telegraph, this site collects data from Web-accessible databases run by the Federal Election Commission, the Yahoo Real Estate database, the U.S. Census and others. By automatically filling out database query forms, a process known as "screen-scraping," Telegraph combines data from disparate sources, yielding the potential to cross-reference names, addresses, individual home prices, crime statistics and other data, leading to ethical questions about the balance between freedom of information and privacy.

* Traffic Flow Visualization: Imagine checking the traffic situation just prior to leaving your office. Unlike most traffic Web sites that present data from a single source, TELEGRAPH assembles data from Web cams, CHP accident reports and loop detectors embedded in the freeways, all on one screen.

Telegraph is known as an adaptive data flow system because it adapts to the fact that data often arrives in unpredictable bursts. Telegraph uses new technologies developed at UC Berkeley to route these sporadic data flows into a steady, manageable stream of useful information. Like the Berkeley street after which it is named, Telegraph is the natural thoroughfare for a volatile, eclectic mix coming from all over the world.