中文摘要：

根据自然语言语义特征提取、匹配的实时性和准确性要求，提出模糊聚类、单亲遗传搜索匹配算法相结合的新方法，通过对候选特征点进行模糊聚类处理，使其分布在高斯差分图像的灰度轮廓线边缘，利用单亲遗传算法找到满足约束条件全局最优语义特征，并把所有语义特征进行分类，给出分类依据．试验证明，此语义特征匹配算法在未知语境环境、语义特征频繁变化的环境具有很强的鲁棒性，能够在自然语言处理过程中实时准确识别段落中的语义特征．

英文摘要：

According to the requirement of real-time ability and validity in feature extraction and matching of semantic feature of natural language, a new algorithm of fuzzy clustering combined with partheno-genetic was presented. The candidate features were processed with fuzzy clustering, so that the extracted features were distributed around gray outline of Gaussian difference image. The globally optimal semantic features satisfying constraints could be found with partheno-genetic algorithm and all semantic features were classified and the classification basis was given at the same time. It was verified by experiment that this algorithm is strongly robust in an unknown language environment and as well in the environment where the semantic feature was frequently changing, so that the algorithm would be able to extract semantic features accurately and in real time from the paragraphs in the process of natural language processing.