基于后缀数组的近似字符串匹配-计算机软件与理论专业论文.docxVIP

I I 摘要 字符串匹配问题是计算机科学研究中最基础的问题之一。早期的研究多集中 于精确字符串匹配领域，就精确字符串匹配提出了许多单模式和多模式匹配算法。 然而在信息检索、模式识别和计算生物学等一些实际应用中有时候更需要查找的 是近似匹配字符串。因此研究高效的近似字符串匹配算法具有重要的理论价值和 实际意义。 本文首先介绍了近似字符串匹配算法的研究现状和近似字符串的相关理论和 主要研究方法—动态规划、自动机、位并行、过滤等技术。结合后缀数组这一索 引结构，本文提出了基于后缀数组的近似字符串匹配算法，即 IS_DP 算法。算法 使用后缀数组加快在动态规划矩阵上的迭代速度，从而达到 O(kn)的时间复杂度和 空间复杂度。用后缀数组构造算法可以预先排序文本串后缀，求解后缀数组中相 邻后缀的最长公共前缀，从而可以在 O(1)时间内计算出任意两个后缀串的最长共 有前缀长度。本文算法是对原有的基于动态规划的近似字符串匹配算法的改进， 通过加速矩阵对角线上扩展的方式降低了构造动态规划矩阵的时间消耗，进而在 O(kn)时间内查找出文本中所有的近似匹配字符串，与一些在后缀树上查找近似字 符串匹配的算法相比较，本文算法采用后缀数组结构可以节省约 5n 的存储空间。 关键词： 动态规划 后缀数组 编辑距离 近似字符串匹配 最长公 共前缀数组 III III Abstract String matching is one of the basic problems in computer science. A lot of researches were focusing on exact string matching in early time, and many algorithms of single and multiple string matching have been proposed. However, in applications such as information retrieval, pattern recognition and computational biology, approximate string matching makes more sense. So there is so much theoretical value and practical meaning to research and design efficient approximate string matching algorithms. In the beginning, this paper introduces the research status and basic theories of approximate string matching, and then presents the main approaches which include dynamic programming, automata, bit-parallelism and filtering. Then we come up with our approximate string matching over suffix array algorithm which is short for IS_DP algorithm. IS_DP algorithm uses suffix array for accelerating the computation along the dynamic programming table and reaches running time in O(kn). Suffix array are used for preprocessing the sequences allowing an O(1) running time computation of the longest common extension between substrings. Compared with the classic approximate string matching algorithm which is based on dynamic programming, our algorithm shows pretty good time complexity which is O(kn); besides, compared with some approximate string matching algorithms which are based on suffix tree, IS_DP algorithm doe