Chinese Patent Machine Translation

Chinese Patent Machine Translation (CPMT)

The Introduction of Machine Translation

In recent years, the booming of patent and IP activities in China has contributed to enormous and urgent demands from home and abroad for machine translation of Chinese patent documents. Serving as a state-level patent information organization, the China Patent Information Center (CPIC) is pleased to announce the launch of a Chinese-English MT system targeting patent documents, which marks a significant breakthrough in this field. The system, developed by the CPIC in collaboration with domestic and overseas companies and organizations, is dedicated to provide users with fullyautomatic on-line Chinese-English MT service.

The Chinese-English MT system for patent documents is designed to be a useful system elaborated to present users with a quick-response and stable environment employing an intuitive interface, which will for the first time offer a full-automatic MT service for patent documents from China through the Internet and which promises a significant promotion of international communication and cooperation on patent documents.

(1) Realizing a multi-strategy method designed as an organic integration of the three MT models, which was developed in orientation to actual Chinese patent documents, thereby providing a pertinent solution to patent-specific linguistic phenomena while ensuring a fundamental consideration on the commonness of linguistic rules, and as a result, promising a sufficient understandability of translated Chinese patent information to non-Chinese speaking users.

(2) Constructing terminology database having a substantially full coverage of technical fields, which enables the translation engine to be adaptive to patent documents featuring a wide coverage of specialties and terminologies, with a scheme explored through analysis and statistics of actual patents, which has demonstrated satisfactory efficiency in terminology construction, as well as in coping with ambiguity and uncertainty commonly existed in MT.

(3) Designing and realizing the hybrid strategy-based Chinese-English MT system for patent documents based on the multi-strategy method. The modularized architecture of the system implementing the design idea and the translation strategies was accomplished through construction of a language model and a language knowledge base, and the accomplished system was seamlessly integrated with an existing retrieval system, functioning with a mechanism of Web service.

State Intellectual Property Office of the People's Republic of China