With the ever growing need of semantically linking data across different domain, organizational and diplomacy boundaries using RDF and OWL, one of the major obstacles impeding the advancement of Semantic Web is the data availability. This paper presents a browser-based semantic mapping tool for converting linked data on the Web to be available for Semantic Web applications. It is aiming at helping users define semantic mapping from relational databases to RDF schemas. This tool provides much easy-to-use functionality to facilitate users in this definition work, such as drag-and-drop mapping, visualization mapping, data source annotation, database server independent, standard mapping language and shared ontology management.
With the fast development of Web, millions of linked document
and linked data are stored in online web applications. As a result,
the Web becomes a global information center with not only linked
documents, but also linked data. In particular, most of linked data
are stored in relational databases that are always adopted as the
data storing space by online web applications. Nowadays, one of
the major obstacles impeding the development of Semantic Web
is the data availability. It is an opportunity to convert these linked
data into Semantic Web as the data sources for semantic
applications. However, one problem is how to integrate
heterogeneous and distributed relational databases. One of the
mature solutions of this problem is integrating heterogeneous data
sources using ontology based on semantic mapping, which are
mappings from database schema to RDF1 [
In this paper, we present a well developed browser-based mapping tool for converting linked data from Web to Semantic Web. Before developing the tool, we verified fundamental model difference between relational schema and the RDF schema and specified a set of mapping issues and mapping cases based on the THALIA2 Testbed. Based on these advance work, we developed this browser-based mapping engine to help users to define semantic mapping. It provides a lot of powerful functionalities to facilitate domain experts in defining semantic mappings. In particular, the tool is able to deal with complex relational schema with complex joins. Besides, it also provides a functionality of transforming relational data to RDF data.
This semantic mapping tool is developed based on the OAT (OpenLink Ajax Toolkit) framework and Virtuoso Universal Server 3 . This tool is a web engine defining semantic mapping from heterogeneous databases to shared ontology namely from relational schema to RDF schema.
This tool aims at helping users to define semantic mapping from relational schema to RDF schema, which needs additional components to support such as data source annotation, shared ontology management, and file management. As the figure 2.1 shows, it is composed of three main components:
Mapping tool for linked data: This component is the main functional part of this tool, which provides much functionality to define semantic mappings from relational schema to RDF schema by drag-and-drop. In the end, it also supports saving the defined mappings as Meta Schema Language, which is defined as script to transform relational data to RDF data executed on Virtuoso Universal Server. Virtuoso Universal Server: This component is charge of interactions with databases, including obtaining Meta schema information, executing queries, and transform relational data to RDF data. 2 THALIA:http://esw.w3.org/topic/TaskForces/CommunityProject s/LinkingOpenData/THALIATestbed
Physical databases: The physical databases stores millions of linked data on the Web.
Browser-based Mapping Tool
for Linked Data
The browser-based mapping tool is well developed to help domain experts define semantic mapping from relational schema to RDF schema. In particular, it provides much easy-to-use functionality and has some special technical features. In this section, we describe these functionalities and features in details.
Figure 2.2 depicts the browser-based mapping tool, including resource views, editing domains, and mapping visualization. It consists of three primary functional components as follows: Building table-to-class mappings: At first, “table” items are dragged from DbResView and dropped under the edit domain; then ontology classes are dragged from OntoView and dropped under the “table” area. Secondly, users drag properties from OntoView and drop them under ontology class, then drag columns of the “table” from DbResView and drop them under the property areas.
Building complex join mappings: Taking table-to-class mappings as the inputs, the tool determines whether there are joins between the tables based on semantic relationships defined in the shared ontology.
Meta Schema Language: This tool use Meta Schema Language as semantic mapping description language. After defining the mapping from relational schema to RDF schema, it provides a functionality to save mappings in Meta Schema Language, which is able to be directly executed on Virtuoso Universal Server to transform relational data to RDF data.
The browser-based mapping tool has many special technical features, which make it outstanding among the other existing semantic mapping tools, such as:
Database server independent – The mapping tool is able to support many usual databases, such as Oracle, Mysql and so on.
Drag-and-drop mapping – Users are able to just drag and drop items to complete the process of defining semantic mappings.
Mapping Visualization – After user dragging and dropping, the items and semantic mappings are visualized and displayed in the edit domain. And this feature makes the engine easier to extend.
Data source annotation – Structures of data sources are displayed in the view on the left side, on the other hand, ontology schema are displayed in the view on the right side.
Semi-automatic mapping for complex join – Based on table-to-class mappings, it can semi-automatically define mappings for complex joins.
4. Indicate the existence of joins bewteen tables 3.Drops tables and classes and establishs their mappings. 1. Display relational tables 2. Display ontologies.
This paper presents a browser-based semantic mapping tool for linked data in Semantic Web. It has been well designed and accomplished to facilitate users in defining mappings. At present, this tool is under testing stage accessed in http://ec2-67-202-2244.compute-1.amazonaws.com:8890/demo/mapper.
In our plan, the tool will be enhanced with some heuristic rules
to automate the semantic mapping task as far as possible, just like
the approach proposed by Yuan An and colleagues [