不完全语义认知过程中信息特征正确识别仿真

摘要

语义信息特征的正确识别能够有效地捕捉虚拟装配过程中用户交互意图.进行语义的正确识别时需要建立语义信息熵和词语权重值之间的关系,确定各个特征的语义权重来完成识别.传统方法利用隐藏在语义的映射关系进行识别,将奇异值分解和遗传思想理论相融合降低向量空间的维数,但没有详细分析信息熵和权重值之间的关系,所以语义信息特征的识别准确率不高.提出一种基于聚类的不完全语义认知翻译中信息特征选择方法.上述方法先依据任意信息义项的语义距离计算出信息语义间的相似度,获取信息词之间的相关度,在此基础上得到不完全语义认知翻译中信息语义扩展度,获取信息词汇间的共现率,对候选词汇集的词依次进行提取,计算信息特征之间的语义相关度,建立语义信息熵和词语权重值之间的关系,确定各个特征的语义权重,计算出每个信息特征在翻译中所提供的信息量,选取信息量较大的作为不完全语义认知翻译中信息特征.仿真结果表明,所提选择精确度高,可以为翻译教学提供了科学的依据.%In this paper,we proposed an information feature selecting method for the identification translation of incomplete lexeme based on the clustering.Firstly,the author worked out the similarity among the information lexeme according to the lexeme distance of any information senses of dictionary entry and acquired the relevancy among the information words.Then,the extended degree of information lexeme was obtained during the identification translation of incomplete lexeme on that basis and the co-occurrence rate among the lexeme words was acquired to extract the word in the candidate word set in turn.Moreover,the lexeme relevancy among the information feature was calculated and the relationship between the lexeme information entropy and word weight value was built to confirm the lexeme weight of each feature.Finally,the information quantity supplied was worked out with each information feature during the translation and the larger information quantity was selected as the information feature during the identification translation of incomplete lexeme.The simulation results show that the method has higher selection.It can provide scientific basis for the translation teaching.