Design Of BiLingual Information Management System(BLIMS) For On-Line Real-time Translation

The research of the thesis is based on the MIIRPS(Multiple Information flow Input Recognition and Processing System).MIIRPS is trying to organically integrate the information processing techniques such as recognition, classification, abstracting, translating, management, retrieval and I/O. Its main goals are â‘  processing the internet information in real-time, â‘¡ integrating the different information processing techniques, â‘¢ accepting and recognizing multiple information sources including pure text, html. OCR, speech and so on, and â‘£ designing the Natural Language human-computer interface.BiLingual Information Management System-BLIMS is one of five sub-systems of MIIRPS while the other four sub-systems are Source Information Flow Input and Recognition System, Internet Real-time Machine Translation System, Target Information Flow Output and Processing System and Natural Language Human-Computer Interface.The main tasks of BLIMS are â‘  auto-classifying and indexing the information, and passing them to the MT system, â‘¡ auto-abstracting and then reprocessing the abstracts for improving the performance of MT system, â‘¢building BiLingual Information Base(BLIB) for storing and retrieving the translated information.There are three parts of BLIMS, i.e. auto-classifying and indexing system, auto-abstracting system, and BLIB storing and retrieving system.The approach based on Hierarchical Dictionary is proposed in auto-classifying and indexing system, which forms the basis of translating, storing and abstracting.Incorporating the experience of the predecessors, a corpus-based approach integrating the linguistic analysis of the text and statistic data is introduced in auto-abstracting system. In addition, the thesis proposes the...