IBM
Skip to main content
 
Search IBM Research
     Home  |  Products & services  |  Support & downloads  |  My account
 Select a country
 IBM Research Home
China Research Laboratory
Research Projects
 ·Speech Recognition
 ·Human Computer Interaction
 ·Natural Language Processing
 ·e-Commerce
 ·Pervasive Computing
 ·Infrastructure Technology
 ·Internet Media
Projects Updates
Visitor Information
Feedback

Related Links
 Careers at Research
 Other Labs
 IBM Journal of R & D
 IBM Systems Journal

Chinese Version
 
 


 
China Research Laboratory
Natural Language Processing

Intelligent information processing

We study how to utilize linguistic and domain knowledge to understand and process unstructured, structured or hybrid information. Our research topics include: named entity recognition, information extraction, summary generation, linguistic search, categorization, clustering, topic detection and tracking, duplicate and similar text detection, etc. The target is to drive these technologies to obtain performance such that they achieve wide scale commercial application. Third-party evaluation shows the accuracy of text categorization has a good lead against the competitors. Currently these technologies are adopted in several enterprise and government intranets or extranets.


Knowledge creation and sharing
We study how to discover linguistic and domain knowledge from massive electronic information or human behavior, as well as manage and use of knowledge among parties. Our research topics include: knowledge representation and knowledge base, ontology discovery, resource distribution, etc. The target is to largely drop the cost of enterprise to acquire and share knowledge so as to make possible to apply intelligent information processing.


Machine Translation
The IBM WebSphere Translation Server (WTS) for Multiplatforms Version 2.0 was announced at the end of March, 2002. This is a machine translation (MT) offering to assist companies overcome language barrier for worldwide communication and business. With WTS 2.0, enterprises can provide multiple language versions of their single source web pages, e-mail messages and chat conversations in real time. Specifically designed for enterprise, WTS can leverage company's existing Web infrastructure and content and provide customer information in various languages, to save enterprise's investment. Sixteen language pairs are supported in WTS 2.0, nine of which are newly added. China Research Laboratory (CRL) developed both English-to-Chinese and Chinese-to-English MT engines inside. CRL also worked jointly with China Development Laboratory (CDL) Taiwan to develop the Traditional Chinese edition of WTS 2.0 for Taiwan and Hong Kong market.


 
  About IBM  |  Privacy  |  Terms of use  |  Contact