摘要
首先介绍采用平滑算法构建通用语言模型、领域语言模型和用户语言模型的方法;然后提出了一种模型融合方法,形成了一个动态自适应语言模型.最后,将此模型与输入系统实现技术相结合,实现了一个基于动态自适应语言模型的中文输入系统.实验表明,应用该模型的输入系统不仅降低了平均码长,而且显著提高了首字命中率.
The paper introduces an approach to construct the general-purposed language model,special-domain language model and special-user language model based on the data smoothing algorithm.Then,it proposes a model fusion method which integrates these models into a dynamic self-adaptation language model.Finally,it implements a Chinese input system with the dynamic self-adaptation language model to verify its performance.The experimental results show that new model not only can reduce the average input code-length,but also can improve the hit rate of the first candidate character.
出处
《苏州大学学报(自然科学版)》
CAS
2011年第2期29-35,共7页
Journal of Soochow University(Natural Science Edition)
基金
国家自然科学基金(90920004
60970056
60873150)
江苏省自然科学基金(BK2008160)
江苏省高校自然科学重大基础研究项目(08KJA520002)
关键词
领域语言模型
用户语言模型
模型融合
动态自适应语言模型
中文输入系统
special-domain language model
special-user language model
model fusion
dynamic self-adaptation language model
Chinese input system