The Uncertainty Of Classification System

China will build a“six unified”natural resource survey and monitoring system,in which the actual integration of existing natural resource survey results is an important issue that needs to be solved urgently^[1].Studying the semantic relationships of categories in different classification systems is one possible solution.Previous research on the interoperating of classification systems is mainly based on category names.However,the semantic information of category definition and the semantic relationships between categories needs to be taken into consideration.This paper introduces the Dynamic Semantic Reference System(DSRS)method to study the semantic problems.Constructing a reference semantic concept tree to qualify semantic information,this paper suggests a new perspective for the interoperation of geographic data.Taking the natural resource survey as an example,semantic problems are summarized and suggestions for building classificaiton systems are given in order to promote the substantive integration of multi-source data,and provide support for building a unified geographic ontology library.The specific contents are as follows:(1)A semantic analysis method based on dynamic semantic reference concept tree is proposed.Firstly,the whole process of constructing DSRS and the definition of concepts involved are introduced,and its methods in semantic analysis of Chinese texts are studied.Then,the method of building semantic reference concept tree is introduced,and the three categories of concepts are taken as examples to elaborate the reference.Finally,the quantification of the category semantic information is implemented according to the reconstruction of the leaf concept in the reference semantic concept tree.(2)Based on the DSRS,three indicators for measuring internal semantic problems,semantic gap,semantic overflow and semantic overlap are introduced and improved.Taking the classification system specified in the Contents and Indicators of Geographical Conditions(GDPJ 01-2013)^[2]and the nature resource suvery of the pilot area^[3]as examples,the semantic problems between categories are analyzed.The experimental results show that DSRS method can discover and locate the semantic problems of classification systems.Based on the analysis of experimental results,four suggestions for the construction of classification system are proposed.(3)From the perspective of the differences between classification systems categories,this paper studies and analyzes the relationship between the natural resource survey classification system and current classification systems(the land use classification system and the geographic country monitoring and classification system).Firstly,the DSRS method is adopted to construct semantic trees,and the calculation methods of semantic transformation degree and semantic similarity index are introduced.Then,three different classification systems are introduced and semantic concept trees are constructed.Finally,the semantic conversion between categories of each classification system is quantitatively calculated.The experimental results show that there are four semantic conversion relationships among the classification system categories.67.5%categories in the natural resource classification system have semantic overlap with the land use classification system,and 72.5%categories overlap with the state geographic monitoring classification system.