Information Analysis & Management

In the era of information society, people are inundated by floods of low-quality, high volumes of information. Nowadays, in most countries of the world, huge numbers of surveillance cameras have been installed in every airport, train station, bank, shopping mall, office building, etc. Each surveillance camera captures every moment of the target scene with a frame rate of 30 fps. Such high-volume, continuous video streams present huge challenges to us because it is unrealistic to rely on humans to monitor such video streams 24 hours a day, 7 days a week without overlooking any important events. The fact is that video streams generated by most surveillance cameras are just recorded on video types or hard disks without being actively monitored.

On the other hand, the advent of worldwide web and blogosphere has dramatically lowered the threshold for ordinary people to publish their own experiences and opinions. This development has lead to the phenomenon that the quantity of information has exploded, while the quality of information has deteriorated. Although Internet search engines can help users to retrieve their desired information through keyword-based search, in general these search engines only take good care of a small number of dominant websites and completely ignore the data in the long tail. However, long-tail data also contain a lot of valuable information, such as grassroots opinions, sentiments, wisdom of crowds, etc. Such information is more valuable for the purposes of business intelligence, better decision making, and market strategies. We need new technologies to dig out such valuation information from the long tail data.

The rapid evolution of web computing has dramatically changed the paradigm of creating and providing online services. In the past, service providers needed to develop every component of a service and secure all the computing resources by themselves. Now they can quickly create new services by taking advantages of the existing web services and developing only those necessary components not available on the Internet. This approach has opened up new business opportunities for service providers to provide specialized, low-cost, high-quality, and fine-grain web services.

In our Cupertino office, we strive to develop cutting edge technologies to dramatically improve humans’ abilities to:

• Sift through large volumes of raw video streams and noisy, low-quality Internet data to extract highly semantic, value-added information


• Summarize and visualize large volumes of Internet data to discover their overall pictures and internal structures


• Mash up heterogeneous information from different sources to create new information services.

By developing these technologies, we want to turn raw video streams into valuable information sources, noisy, low-quality Internet data into knowledge-bases, and complex web service creations into simple cut and paste operations with minimum SI requirements. Such technologies have great potentials in a variety of areas such as intelligent video surveillance, customer attributes and shopping behavior recognitions for retail business, targeted advertisements, grassroots wisdom discovery, market analysis, business intelligence, low-cost, rapid web service creations, etc.