语句级汉字输入技术

本文讨论了包括声音输入、键盘输入、文字识别等各种形式的汉字输入技术的研究和发展, 阐述了按照字、词、语句作为汉字输入技术发展阶段的思怒, 提出了适用于上述各种形式的类码语句歧义处理问题, 该问题可描述为有向图求最短路径的问题。本文讨论了采用语法—语义分析和统计模型的最少元素概率推理方法和控制策略, 在知识库完备或不完备的情况下均可进行正常的推理, 并给出基于当时情况下的最佳结果。本文还简要介绍了几个应用事例。

Chinese Input by Sentence

This paper discusses the research and development of Chinese input technology in keyboard, speech and character recognition fields, which can be dividing as three stages:character-input, word-input and sentence-input. In the concept of a directed graph, a method is presented to solve the main problem of the third generation Chinese input system, i. e. the problem resulted by different meaning of the sentence inputted by keyboard or speech system or recognition system. Various experiments show that higher flexibility and precision have been achieved by the method. The typical real sentence-input systems are introduCed also.