用於鏈結資料上引導SPARQL的REST風格網路服務

Abstract

近年來由於資料量急遽增加，如何儲存這些資料並有效運用成為目前面臨最大的挑戰。在”Web 3.0”的世代，語意網是其中一項特徵，我們希望讓資料與資料之間多一層”語意”的關係來形成鏈結資料，讓資料更容易被機器或是人運用，但如何讓使用者都能容易使用並加上這一層”語意”關係即為最大的問題。 我們提出一個REST風格的網路服務，目的在於讓不會SPARQL的人也可以使用鏈結資料並且容易發展及應用在多個平台上，如Android手機、入口網站……等。如何將REST應用程式介面轉換成能夠操作鏈結資料的語言和如何利用Ontology幫助使用者建立資料間的關係即為此篇論文重點。在此篇論文我們討論了：（1）如何把SPARQL形式轉換成REST的請求方法、（2）如何從SPARQL對應到REST的資源、（3）如何從REST應用程式介面得到的資料並轉成SPARQL。此種將REST應用程式介面轉成SPARQL的方法，我們分成兩個方面來考慮，在查詢方面，我們利用圖的資料結構和圖的演算法，首先讀取Ontology、把Ontology轉成圖的資料結構，在使用圖的演算法找出最短路徑，並以此路徑來生成SPARQL；在新增/刪除/更新方面，使用REST應用程式介面來新增/刪除/更新一個鏈結資料的Individual。 此系統已使用在我們實驗室發展的BCI Ontology上，並且利用REST應用程式介面也已實作出兩個應用程式CerebraApp和CerebraWeb，分別是Android應用程式和入口網站，期許這套架構能讓鏈結資料發展更加蓬勃。

In recent years, data are growing rapidly. We are now facing the biggest challenges about how to store these data and use them effectively. In the generation of “Web 3.0”, Semantic Web is one of the features. We hope there are another “semantic” relationships between two data and use the relationships to form Linked Data in order to make the data more useful for the machine or human being. However, how to make it easy for people to build the semantic layer is the critical problem. The reason why we propose a RESTful Web Service is that because we hope the people who don’t know how to program SPARQL can also use Linked data. Then, Linked data can be used and developed easily onto any platform, like Android, Web Portal, … and so on. The point of this thesis is how to convert REST API into Linked Data language and how to use Ontology to build the relationships between two data. In the thesis, we discuss: (1) how to map SPARQL form into Request Method of REST. (2) how to map SPARQL into REST resources. (3) how to map REST API into SPARQL. We can divide the approach that maps REST API to SPARQL into two concepts. In the query form, we use the graph data structure and graph algorithm. At the beginning, we load Ontology, transform it into graph and store it into graph data structure. Using the algorithm of graph is the great method to find the shortest paths of the graph. Finally, we can use the paths to generate SPARQL. In the update form about create/delete/update, we use REST API to post/delete/put an individual of Linked Data. This system has been used in BCI Ontology which is developed by our laboratory and there are two applications using the REST API. One of the applications is developed in Android and the other one is developed in JSP to make Web portal. We hope this system architecture can make Linked Data more vigorous.