人机交互中语义理解和一致性维护问题的研究

The bottleneck of input and output between computer and human is an obstacle for the application of computer. In order.to promote the application of computer, the idea of natural interface between human and computer is introduced, i.e., the computer can speak and listen to people as human so that the interaction between human and computer is smmothly as that between human and human. The machine learning and knowledge base process technology are applied to this kind of interface, and the knowledge base in the interface is dynamic. The knowledge base.should be revised as the new rules are learned. We propose a model for this kind of interface which can learning and revising its knowledge base. Several approaches are introduced to revise the.knowledge base in a interface, and applied to diagnosis. A knowledge-based approach to multimodal intergeration is introduced.

给出人机交互中知识库一致性维护的可实现方法，以及一些特殊知识库一致性维护的高效算法，实现一致性维护的原型系统；给出适合上下文的表示方法和推理机制，建立上下文的硎竞屯评淼幕纠砺郏谎芯坑镅缘挠镆謇斫猓瞿承┨厥饣肪持械挠镆謇斫夥椒ǎ越饩鋈嘶换ブ械挠镆宸蠢∥侍狻８醚芯课匀蝗嘶缑嫣峁├砺刍『图际踔С帧