Abstract: In order to construct phylogenetic tree with missing morphological data, this paper proposes an attribute reduction method for constructing phylogenetic tree. Firstly, both prior knowledge and more complete data are employed to construct an initial phylogenetic tree, Then, attribute reduction strategies are applied to get the decision nodes, and the decision model are constructed based on the decision nodes. Finally, the position of the species with high proportion of missing data in the initial phylogenetic tree is determined by the constructed decision model, and the phylogenetic tree is constructed using species grafting technology. The comprehensive experimental results show that when the proportion of morphological missing data of a single species is greater than 10%, the average species accuracy with the proposed method is about 10% higher than the Maximum Parsimony（MP）.

Key words: phylogenetic tree construction, morphological missing data, attribution reduction, prior decision model

摘要： 为了解决含有缺失形态学数据谱系树的构建问题，提出了运用属性约简构建谱系树的方法。首先，利用先验知识和较完整的部分物种数据构建初始谱系树；然后，运用属性约简原理获得属性决策组集合的决策点，进而建立先验决策模型；最后，根据先验决策模型确定缺失数据比例较高的物种在初始谱系树中的位置，通过物种嫁接完成谱系演化树的构建。实验结果表明，当单个物种缺失数据比例大于10%时，相比最大简约法在平均准确率方面平均高出10%左右。

关键词: 谱系树构建, 形态学缺失数据, 属性约简, 先验决策模型