中文摘要：

词义消歧解决自然语言中同形异义词语在不同上下文环境中的义项标注问题,是自然语言处理领域的基础性关键问题.核方法是机器学习中一类强有力的统计学习技术,被广泛应用于分类、回归、聚类等诸多领域.基于核方法的词义消歧的关键是如何构造一个能够充分表达待消歧词上下文信息的核函数.在介绍基于核方法的词义消歧系统的一般框架之后,系统阐述了国内外面向统计词义消歧的核函数构造与选择的研究现状及进展,重点分析了研究中存在的问题及解决方法,最后探讨了未来研究的重点与可能的发展方向.

英文摘要：

Word Sense Disambiguation（ WSD） aims to determine the most appropriate sense of homographs in different contexts and is one of the fundamental key issues in natural language processing. Kernel methods are a class of powerful statistical learning technologies in machine learning,which have been widely applied in classification,regression,clustering and so on. The key to the success of the kernel methods- based WSD is to construct an appropriate kernel function that can fully convey the information of the contexts of the word to be disambiguated. After giving an introduction to the generic WSD system architecture based on kernel methods,we systematically surveyed the research state and advances in kernel construction and selection for statistical WSD,concentrating on the existing problems and their solutions. Finally,we discussed the research emphases and possible directions of the future work.