基于城市地址树的地址文本匹配方法

为了适应中文地址数据的复杂性,本文依据其中地址要素的层级关系,建立城市地址要素的树形模型,并提出基于地址树的文本自适应匹配方法,该方法根据地址数据中各部分地址信息匹配的节点评价选择最优匹配结果,通过单元最大长度匹配法获得地址树中与地址信息相匹配的节点,参照节点的层级关系构建相对独立的地址节,根据地址节中的地址信息计算权重因子,回溯评价返回最优匹配结果;本文采用深圳市518 948条建筑物地址数据构建城市地址树,在此基础上进行地址匹配试验,达到85.6%的匹配准确率,可应用于地址标准化和地址匹配流程中。

Address Text Matching Method Based on City Address Tree

Based on the hierarchical relationship of Chinese address element in the tree model city address elements,text matching method is proposed in this paper based on the adaptive address tree,according to the evaluation data of node address information to achieve the optimal matching result.Through the matching method to obtain the maximum length of unit node address tree and address information matching the hierarchy reference node,this method establishes a relatively independent address section,the weighting factor is calculated according to the address information address section,backtracking returns the optimal matching evaluation results.Shenzhen City address tree is constructed with 518948 address records,based on this ,the address matching test achieved a good matching accuracy.