IDG News Service - Microsoft researchers in Beijing are developing applications that mine online data to track human relationships and help with translation, lab managers said Monday.
Another program in development analyzes satellite positioning data to direct users to interesting locations by mobile phone.
The lab, Microsoft's research base in East Asia, has produced one application that maps a person's connections to friends and colleagues when a user searches for the person's name, said Wei-Ying Ma, assistant director at the lab. Clicking on the line between two people gives a pop-up summary of their relationship.
The application, called the Entity Cube and currently available only in Chinese, creates the maps based on public online information. Its database draws on archived Web content to determine the type and the strength of each relationship, and Microsoft computers crawl 200 Chinese news Web sites to update it as new information hits the Internet, he said.
The lab plans to add Twitter messages to the database. Possible future sources of updates include travel sites, social networking sites or instant messages sent via programs like Windows Live Messenger, said Ma. People could also be allowed to add information about themselves to make the database more complete, he said.
One problem for the search engine is potential mixing of information when people share a name, but it can sometimes distinguish between people if they often appear in separate contexts, such as two different businesses, said Ma.
The application could eventually create revenue through advertising or by offering analysis of its data on people and firms, Ma said.
"Right now, it's still used to improve search, but we believe all the knowledge we extracted from this cloud data can enable a lot of... Internet services and applications," Ma said.
The Beijing lab is also developing a mobile application that could suggest nearby sights for travelers or stores for shoppers.
The program, called GeoLife, draws on GPS (Global Positioning System) information gathered from third-party vendors to make its suggestions. The lab's computers mine that data to find, for example, locations that are not just close together but also offer similar attractions or services, said Ma.
The next step for both the relationship search engine and GeoLife is to attract users, whose behavior and input will expand the databases for the applications, Ma said.
The Beijing lab has also produced a Chinese-English dictionary that searches the Web to find and absorb new words, phrases or pictures, said Hsiao-Wuen Hon, the lab's director.
The dictionary can suggest corrections for wrong word choices as Chinese users type in English. Its ability to draw endless new information from the Internet makes it perhaps the most extensive bilingual dictionary available, and it can serve as an English thesaurus as well, Hon said.
Free Insider Guide
Instead of bulk capital expenditures for large servers and storage arrays, you can purchase computer time based on actual usage of CPU cycles and storage by the number of gigabytes or terabytes used. But here are ten things to consider before you jump into the cloud. Insider (registration required) more
Copyright © 1994 - 2012 Computerworld Inc. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of Computerworld Inc. is prohibited. Computerworld and Computerworld.com and the respective logos are trademarks of International Data Group Inc.
