Abstract

XBRL(eXtensible Business Reporting Language)is a kind of extensible marked-up language which faces financial and political fileds.It has been used so widely that XBRL financial reports are increasing repidly which is of great significance to data storage and analysis of XBRL files.However,the logical structure of XBRL files are quite complicated and lacks description of data obejects.Therefore,existing data storage and analysis tools can not accomplish analysing job. This thesis will do some researches on XBRL data storage and solve this problem.
This thesis firstly does deep researches on XBRL and puts forward a XBRL data storage schema based on CWM(complete warehouse metamodel).This idea comes from the newly import of XBRL dimesions2.1 standard.It not only solves the conflict between flat relationship database and multidimensional structure of XBRL,but also realise the integration on meta-data level of XBRL files and data storage and analysis tools.So the new model is independent,universal and extensible.Based on this,the thesis imports prefix encoding and analysis method improvement to improve query efficiency aiming at the high efficiency requirement of storage engine module.This thesis proposes a new encoding method for XBRL based on prefix encoding and redundancy concept.This new encoding method will both keep the advantage of high query efficiency of prefix encoding and reduce XBRL subtree update costs by setting reserving room.In order to increase parsing efficiency of XBRL data, this thesis proposes a combination method of DOM and SAX, which proves to be effective. 
Finally, this thesis designs and implements a XBRL data storage system based on CWM.This system contains two modules:data storage module and data engine module.The storage system can show XBRL data in a multidimensional model and is compatible with data storage and analysis tools supporting CWM standard, which is sure to have a bright future.

key words:XBRL;CWM;multidimensional model;