Some new concepts of effective incidence matrix,ascending order adjacency matrix andend-result vertex are introduced,and some improvements of the maximum weight matchingalgorithm are made.With this method a computer p...Some new concepts of effective incidence matrix,ascending order adjacency matrix andend-result vertex are introduced,and some improvements of the maximum weight matchingalgorithm are made.With this method a computer program in FORTRAN language is realized onthe computers FELIX C-512 and IBM-PC.Good results are obtained in practical operations.展开更多
正向最大匹配分词FMM(Forward Maximum Matching)算法存在设定的最大词长初始值固定不变的问题,带来长词丢失或匹配次数较多的弊端。针对此问题提出了根据中文分词词典中的词条长度动态确定截取待处理文本长度的思想,改进了FMM算法。与...正向最大匹配分词FMM(Forward Maximum Matching)算法存在设定的最大词长初始值固定不变的问题,带来长词丢失或匹配次数较多的弊端。针对此问题提出了根据中文分词词典中的词条长度动态确定截取待处理文本长度的思想,改进了FMM算法。与此相配合,设计了一种词典结构,使之能够有效地支持改进的算法。改进的算法与一般正向最大匹配算法相比大大减少了匹配次数,分析表明中文分词的速度和效率有了很大提高。展开更多
文摘Some new concepts of effective incidence matrix,ascending order adjacency matrix andend-result vertex are introduced,and some improvements of the maximum weight matchingalgorithm are made.With this method a computer program in FORTRAN language is realized onthe computers FELIX C-512 and IBM-PC.Good results are obtained in practical operations.
文摘正向最大匹配分词FMM(Forward Maximum Matching)算法存在设定的最大词长初始值固定不变的问题,带来长词丢失或匹配次数较多的弊端。针对此问题提出了根据中文分词词典中的词条长度动态确定截取待处理文本长度的思想,改进了FMM算法。与此相配合,设计了一种词典结构,使之能够有效地支持改进的算法。改进的算法与一般正向最大匹配算法相比大大减少了匹配次数,分析表明中文分词的速度和效率有了很大提高。
