摘要:
本文运用标记法理论讨论和处理了计算机化学结构信息处理中的化合物子结构识别问
题.文中首先给出并定义了属性、互协调标记对、全协调标记集等一系列术语和概念;重新定
义了标记法中原已有的约束关系集、相容性、协调性等概念;指出了化合物子结构匹配等价于
全协调标记集的求取.在此基础上引出了求取全协调标记集的有效算法.最后,对随机采集
的近两千例样品进行了试识别,并与Ullmann[2]的算法作了比较,结果是肯定的.
Abstract:
In this paper, the problem of chemical substructure matching is handled within the frame-work
of labeling method. First, some definitions such as attributes, mutually compatible
labeling pairs and completely compatible labeling sets and so forth are given; some concepts
such as constrain relation sets, consistency and compatibility, which have appeared in the
early labeling theory, are redefined. And the equivalence between monomorphism and the
achieving of completely compatible labeling sets is demonstrated. Then, an efficient algorithm
to deal with this problem is provided. Finally, two thousand test matches, collected randomly
are carried out to compare this algorithm with Ullmann's. The results obtained are
satisfying.
下载:
