本研究將可延伸性企業報導語言(eXtensible Business Reporting Language, XBRL)轉換為鏈結資料(Linked data, LD)形式,亦即建立可鏈結化的XBRL,並與其他鏈結資料進行鏈結,讓鏈結化XBRL成為網路共用共享的資料,即所謂web of data,以整合財務報導及其他相關資料等異質資料。XBRL格式的企業報導,其資料散布在不同的檔案中,而運用語意網技術得以將資料結構統一,並能運用SPARQL查詢取得資料,達成異質資料整合,及跨資料資訊查詢,因此運用語意網技術可以整合XBRL及其他財務相關資料,且易於查詢。本研究依據過往文獻指出目前資料無法識別、缺乏鏈結、及不具語意三點,以Tim Berners-Lee提出的開放資料五星評等中第五星評等條件及四項鏈結資料建置原則發展鏈結資料解決問題,首先建立詞彙概念架構並轉換成RDF形式開放資料;其次則是定義資料概念間的邏輯關係;最後則是混搭資料進行SPARQL查詢,獲得跨資料的資訊。本研究實驗結果顯示,利用前述的鏈結資料建置程序,能夠完全符合五星評等中第五星的條件,讓資料成為網路上共用共享的資料;此外利用混搭鏈結化XBRL,並運用SPARQL查詢語言進行查詢,較以往做法更易於整合異質資料,比起現有的資訊揭露網站,鏈結化XBRL的資訊完整率達80%,而現有其他網站的資訊完整率僅有60%。因此本研究所提出的鏈結資料建置程序能夠使資料成為鏈結資料,而本研究所提出的鏈結資料混搭方法能使鏈結資料易於找到其他可進行互補的資訊,使資訊的完整率提升。
This study converts eXtensible Business Reporting Language (XBRL) to Linked Data (LD) to enhancing the usability of XBRL data on the Web environment. The enhanced XBRL (so called LD-XBRL) data inherits the features from both XBRL and LD, such as reusable and shareable. In this study, we focus on the sharable capability of LD-XBRL data. LD-XBRL data is a member of “web of data” that can either share itself to others or mashup data from others’. Using semantic web technique can help achieve the unification of data structure and the integration of heterogeneous data by SPARQL, and SPARQL has ability to require data by querying cross-database information. Tim Berners-Lee proposed the four rules of building linked data and 5-star ratings standard of open data. According to the past references, there are three problems which lack of identify, linked and non-semantic become non-reusable and unshareable data. In this research, we proposed to solve above problems by 3 steps which based on the 5th star open data. First, build the conceptual terms framework and transfer into RDF-Based open data. Definite the logical relationship between conceptual data in the second step. Finally, use the query language of SPARQL to mashup data. This research of experimental results show that utilizing Linked Data Build Process could absolutely accord with 5th star condition, also it can help become reusable and shareable data on internet. In addition, using mashup Linked XBRL has over 80% accuracy rating, and it is higher than other website information completion rate to 20%. As a result, this research proposed the Linked Data Build Process could let separate data become Linked data, the Linked Data Mashup Method could help linked data find other complementary information to improve information completion rate.