完全派生地名和通名派生地名的挖掘  

Mining of completely derived toponym and generic derived toponym

作　　者：刘汉优 王继周[1] 毛曦[1] 马维军[1] LIU Hanyou;WANG Jizhou;MAO Xi;MA Weijun(Chinese Academy of Surveying and Mapping,Beijing 100036,China;School of Geomatics,Liaoning Technical University,Fuxin,Liaoning 123000,China)

机构地区：[1]中国测绘科学研究院,北京100036 [2]辽宁工程技术大学测绘与地理科学学院,辽宁阜新123000

出　　处：《测绘科学》2022年第10期176-181,220,共7页Science of Surveying and Mapping

基　　金：中国测绘学院研究院基本科研业务经费项目(7771802,7771721)

摘　　要：针对全球地名数据中缺少派生地名的标注以及相应原生地名、派生类别、位置关系等信息,造成派生地名翻译、派生地名研究与回溯障碍以及人工甄别完全派生地名和完全派生地名并标注派生信息效率低的问题,该文提出了基于GeoHash编码的通名派生和完全派生地名识别算法。首先,利用地名地理坐标数据进行GeoHash编码并创建GeoHash地名字典,再筛选出疑似的通名派生地名和完全派生地名;然后,利用基于GeoHash编码的邻近地名搜索算法获取邻近地名并进行原生地名的筛选;最后,计算地理关联度,进而根据相对位置、派生规律、地物类别关系、地理关联度4个约束条件实现完全派生地名和通名派生地名及其原生地名的识别,从而获取派生信息。实验结果表明,在给定区域内的地名数据中,该算法能识别出大部分存在通名派生关系和完全派生关系的派生地名及其原生地名。Aiming at the lack of information such as the labeling of derived toponyms and corresponding original toponyms,derived categories and positional relations in the toponym data,which causes obstacles to the translation of derived toponyms,the study of derived toponyms and the tracing of derived toponyms,and the problem of time-consuming and inefficient to distinguish fully derived toponym and general name derived toponyms and labeling derived information manually,a GeoHash coding algorithm for the recognition of general name and fully derived place names was proposed in this paper.Firstly,geographic coordinate data of toponym were used to conduct GeoHash coding and create GeoHash toponym dictionary,then the suspected general name derived toponyms and fully derived toponym were screened out.Secondly,adjacent toponym were obtained and selection of native toponym were conducted by using adjacent toponym search algorithm based on GeoHash coding,then geographical correlation degree were calculated.Finally,according to the four constraints of relative position,derivation rule,ground object category relation and geographical correlation degree,the recognition of fully derived toponym,general name derived toponym and its original toponym were realized and the derivation information were obtained.Experimental results showed that the algorithm could recognize most derived toponyms that have fully derived relation or general name derived relation and their original toponyms.

关 键 词：通名派生 完全派生 邻域查找 最大似然估计 地理关联度 正点互信息 

分 类 号：P208[天文地球—地图制图学与地理信息工程] TP391.1[天文地球—测绘科学与技术]