归一化Alignment距离

Normalized Alignment Distance

为了更好地度量不同序列之间的差异,本文在给定度量空间上的符号距离满足特定条件的前提下,以Alignment距离为基础,提出了一种归一化Alignment距离.接下来利用Alignment相关度证明了该距离满足度量定义的三个条件,并且取值在[0,1]区间上.最后对其进行了进一步讨论,从理论上说明该定义的合理性.该距离可以在序列比对、聚类分析以及模式识别等领域中发挥重要作用.

In order to better measure the difference between the various sequences, in this paper a concept of normalized Alignment distance based on the Alignment distance is proposed when the distance between two symbols in the given metric space meets certain criteria. We also prove that the new distance satisfies the three conditions of the metric definition by using the Alignment relevance, and show that the value of the new distance is in the interval [0, 1]. Finally, further discussion is presented. The reasonability of the proposed definition is evalu-ated theoretically. The proposed distance can play an important role in various practical areas such as sequences alignment, cluster analysis and pattern recognition.