Database Schema Integration For Interoperable Gis Application

Transcription

1 Database Integration For Interoperable Gis Application Norayu Abdul Ghani Nanna Suryana Safiza Suhana Kamal Baharin ABSTRACT This paper discusses database schema integration to support interoperability of GIS applications. Different GIS applications create multiple type of database representation. Rapid development of internet technology allows more spatial and tabular data available and accessible widely across local state, federal and private data clearinghouse. Unfortunately those data cannot be utilized directly due to the differences in data acquisition technique, data definition and their semantic meaning; different coordinate system yet operated within different GIS platforms. As spatial data become core part of business and decision making, therefore data sharing is driven by the need to maintain more accurate and up to date spatial databases and at the same time reduce the data acquisition and maintenance costs. This shows the need of interoperability among different GIS platforms. One of the most difficult and challenging task is to identify and determine automatically how the data from one application matches semantically with the data from the other applications. The ongoing research is intended to provide an alternative solution to solve problems associated with semantic aspects of interoperable GIS. Ontology relationship approach allows query translation from one database schema into another one. Within the whole framework of the ongoing research, a method for database schema integration to support the interoperability of GIS been proposed. The discussion focuses on building conceptual mapping as a prerequisite in developing XML base document. This work hopefully contributes to the information sharing establishment among diverse organizations with heterogeneous GIS database schema. Keywords Interoperability, Database, Semantic, XML Technology, Ontology. 1. INTRODUCTION This research been conducted under the framework of GIS Interoperability Research Project financed by UTeM s short grant and e-science Fund. This project is intended to increase the usefulness of GIS in different possible application domains. Within last few decades, GIS been accepted as a prevailing tool for scientific investigations, resources management and development program. In this regard, GIS shows its strong capability for assembling, storing, manipulating and displaying spatial relationship [1]. However one of the primary obstacles in GIS application is the heterogeneity of the GIS data sources. As mentioned in [2] the heterogeneities can be classified as syntactic, schematic and semantic heterogeneities. From observation it becomes apparent that rapid development of Internet technology support different data and information to be easily available and more accessible widely and freely. However those data are not always useable for other users due to lack of interoperability and compatibility. In addition to this, information sharing between two or more GIS platforms have not been fully implemented. For this reason, unnecessary redundant spatial data acquisition conducted by different GIS users becomes unavoidably occurred. Data acquisition is usually very costly and time consuming. Several vendors have introduced export and import conversion machine. However it been found as the main cause of losing too much data and accuracy. Editing spatial data is also problematic [3]. Therefore issues related to interoperability of GIS have been discussed by millions of GIS users all around the world. There are many research issues related to GIS interoperability. DB A Ontology, XML Federated Federated DB Figure 1. Integration Framework. DB B As presented in Figure 1 this paper discusses the integration of two different database schemas to produce a common interoperable database format. The problems associated with heterogeneity of databases are generated by the use of different data models, schema and data. Brief explanation as above shows the great need for interoperable GIS which been considered to become the best option of the current and future GIS architectures. The introduction of this promising technology will be economically beneficial to all GIS users those who have plan to utilize the existing available data. 311

2 In addition to the above proposed solution a uniform data access to the different data sources is also required. Figure 2 below shows the nature of existing information. Figure 2 (a) illustrate pieces of information could have different format, from different source and varying in quality. complement each other: by extending a simple model, it can be applied it to many different domains [7]. It is a metalanguage, and thereby it must be the best where the task of standardization is concerned. In addition to that, XML proves to be a very efficient format in which the textual data can be published and exchanged. Bibliographic records rarely include anything but text, and they are often very large in size. a b 3. DATABASE SCHEMA INTEGRATION CONCEPT: SIMILARITY AND DIFFERENCES Based on integration framework, Figure 3 below explains that the ultimate objective of database schema integration is to create one common format of database which allows different users to relocate and use the required geospatial data. c Figure 2. Information Integration (a) Piece of Information (b) The ideal situation (c) The Reality. Figure 2 (b) shows the integration of all these pieces of information should construct a compatible format with compatible semantic. However the reality is some points of these pieces are not always compatible with another as shown in Figure 2 (c). 2. XML, GML and RDF Extensible Markup Language (XML) [4] is a general purpose markup language that allows its user to define their own tags. Geography Markup Language (GML) [5] is the XML language for geography developed by the Open GIS consortium. Whereas the Resource Description Framework (RDF) [6] is a framework designed as a metadata model. RDF schema can be used to express typing information. RDF based approach to geography allows free composition between other RDF based languages. Geographic statements can be mixed with statements about weather, physics, business processes, weblogs and syndication, genealogy, politics, and so on, without the need for prior coordination between the designers of languages for these domains. Compared to basic XML, RDF is simple. Unlike the order of RDF properties, order of elements in XML does matter. XML could provide all needed structure for metadata. GML on the other hand is not directly compos able with other XML languages. It is not possible to inject information about geographic features from another application area (geology, crime statistics) into the innards of a GML file, and expect any of the tools to keep working. The choice for XML as main technology for database schema integration depends on the function of XML which capable to define other mark-up languages. XML a very simple model. Essentially, simplicity was a key requirement in the development of XML. Several of the explicitly stated development goals in the XML specification directly support this quality. The specification also states that XML shall be generic, i.e., that "XML shall support a wide variety of applications." Simplicity and generality Figure 3. Common Format of Database. This paper proposes an XML-based solution to the problem of spatial information integration. It is considered as a common interoperable format. The focus of this paper is on how to integrate different schema from different GIS application databases associated with different GIS users and vendors who have defined many application-specific data models. Object oriented DBMS User Interface Federated Import Export Relational DBMS Common Format WEB File System Provided by Provided by Clearinghouse Provided by Translation and Transformation Function Provided by Data Source Heterogeneous Sources Figure 4. Architecture of Integrated Database. In the transformation and translation of source data involves not only earth-referenced spatial data between source data and targeted data in different computer systems, but includes the 312

3 spatial data attribute, geo-referencing, data quality report, data dictionary, and other supporting metadata. Figure 4 above shows architecture for integrated database schema. As presented in Figure 4 and stated by [7], the Federated Database consists of a collection of possible heterogeneous, interoperating but autonomous component databases. It become the preferred strategy for reconciling the proliferation of private and independent databases, and for increasing the productivity of information technology investment. Based on how Federated Database is defined as above, it becomes apparent that traditional searching based on keyword is not sufficient capacity to check semantic meaning of searched objects since it only considers them as character strings [8]. From ongoing research and also supported by [10] it is clear that the key points for a semantic search refinement process depend on the availability of domain ontology and the ability to understand semantic relationships between ontology concepts as explained in Figure 5. Water : Ψ1 Ontology: β1 Wave Length Stream : Ψ2 Ontology: β2 Figure 6. Different Ontology Describing one Entity. Therefore when two classes in two databases schema definition are referring to the same concept, then one global class definition must be created to represent two classes in the component schemas. For an example in Figure 6 is generalized water body super class. Figure 7 shows the global class definition of water body inherits two local specialized class definitions river and lake with their associated attributes namely distance, height and area respectively. Water Ψ 1 β 1 β 2 Ψ 1 (β 1 ) Ψ 2 (β 2 ) Figure 5. Basic Rules for Merge of Ontology. Ψ of set of information denoted as Ψ 1,.,Ψ n and set of Ψ of set of information denoted as Ψ 1,.,Ψ n and set of ontology β for schema of set of information denote as β 1,., β n, Merging ontology, β is based on finding similarities or differences among schema of set of information, Ψ. Giving that two ontology β 1 and β 2 from two different schema Ψ 1 and Ψ 2 respectively, the similarity could be defined as where (Ψ 1 Ψ 2 ) or differences as (Ψ 1 / Ψ 2 ) for these two schemas. Based on the basic rules, thus above schemas could have (Ψ 1 (β 1 ) Ψ 2 (β 2 )) to present the intersection of the ontology of two schemas and (Ψ 1 (β 1 )/Ψ 2 (β 2 )) as shown in the equation 2.1 and 2.2. Integration between two or more databases requires one global definition to represent two different classes within two different databases schema but represent the same concept. This is possible since semantic model allows to present data in a very abstract and understandable manner. It been currently used in designing the conceptual structure of database. When one source of data could be direct transform and translate to desire targeted data, this could eliminate duplication at the same time avoid problem of multiple updates, and minimize inconsistencies across applications. Ψ 2 Stream Wave Length Figure 7. Merging by Finding Equal Concept. Semantic definition here is mapping between an object modeled, represented and/or stored in an information system and a real world object(s) it represents. This mapping represents the semantic of the object being modeled by describing or identifying the meaning and the user perspectives. Generally ontology concerns about what kind of things exist what entity (real things) they are in universe; meanwhile in information technology, ontology can be understood as description of the working model of entities and interaction in some particular domain of knowledge or practices. In general, ontology in this work could be understood as the specification and conceptualization used to help program and human to allocate knowledge in order to generate an agree upon vocabulary for exchange of information As discussed in [12] three main methods could be applied to develop the ontology, and the component of the ontology integration system consists of global ontology, local ontology, and the mapping between local and global ontology. Ontology with type of multi-value attributes may connected by binary, symmetric, many to-many roles while attribute and roles can be inherited through inheritance (is-a) links between classes. The semantic mapping could be broken into three pe namely (1) Identifying the concept (2) Brainstorming (3) Categorization. 313

4 Watershed Groundwater Waterbody Unit Hight Surfacewater Name Figure 8. Common Ontology for Water Body. An example Figure 8, we identify river which is a part of water system as a concept to be semantically map. The next pe is an attempt to explain how we could integrate new information with our existing framework of knowledge. Categorization is to identify the relationship among the new information to the existing knowledge to form the above schemata map. Two semantic relation play important role in the specification of ontology called is-a relation and part-whole relations. relationship is to indicate that one class is a subclass to another class. Part-whole relationship indicates that one or more of the object is part of another object. Ontology allow user to convey queries in their own terms according to their own conceptualization without having to know the underlying modeling and representation of data in heterogeneous databases. Concept used by the user in a query can be then compared in order to search not only for what the user explicitly requested but also for semantically similar terms. These concepts are compared at the ontological level where there is a more complete description of the semantics of terms. Saying that two entities are similar means nothing, since similarity cannot express categorization unless we understand how the similarity is processed with respect to what property or properties they are similar. Using set theory, similarity [13] S, is measured in term of a matching process; this measurement produces a similarity value that is the result of common as well as different characteristic of class. It is calculated as a function of common and different features as S ( ). In S ( ), and are two entity classes, and correspond to the description sets of 1 and, and is the first class that subsume and by the is-a or whole-of () relation. S ( ) (2.1) 4. XML TECHNOLOGY As presented in Figure 9, the XML technology offers significant advantages in the data exchange between interoperable systems due to its flexibility and richness in data representation. Translation and transformation enable source data to be converted to target data as desire by the requester with correct semantic meaning. Since GIS spatial data are stored in hierarchy database, Object relational mapping among database is more suitable. This could model XML document as a tree of objects that are specific to the data in the document. Since XML enable cross platform communication, to achieve interoperability we need such a method to standardize the service request and response into an XML-based document that will be exchanged among service providers and service requesters will significantly improve the communication and implementation process for interoperable GIS services. As mentioned there are so many software and hardware systems in market employable for GIS users to represent and model certain types of phenomena they created. Thus we need technique to represent all the schemas exist into one command way that enable everybody to share those data. Mapping is command as a basis for transferring database to XML and from XML to database. Two type of mapping are possible, table-base mapping and object-relational mapping. However object-relational mapping is more suitable due to capable handling huge subset of XML document. Spatial Data Platform dependent Translation and Transformation XML File Platform Independent Figure 9. Spatial Data Transformation and Translation to XML File. Section bellow provides overview presentation of XML schema, Object Oriented Database (OOD) schema and Relational Database Management System (RDBMS) schema. Figure 10 - Figure 11 below and Figure 12 below shows the schema for object defined in Object Oriented Database and Tables based on merging ontology as in Figure 7 above. Where (2.2) Objects Object Water { =, = = wave Length } Tables Tables A Wave Length a b ; β1 and β2 is term of comparison where β1 is the target and β2 is the base. Cardinality is denoted as and α is a function defines the relative importance of non common characteristics. Figure 10. s for Merging Ontology (a) OOD (b) Table. 314

5 XML <Water> <>, </> <> </> <>Wave Length</> </Water> Figure 11. of XML for Merging Ontology. Class Water { String, String, String, } Create Tables Water ( varchar (10) not null varchar (10) not null varchar (10) not null ) <! Element Water (,, Water)> Figure 12. Mapping between DTD, Classes and Table. OOD and Table are able mapped to XML document. Correspondingly there is apparent mapping between Document Type Data (DTD), Classes and Table as show above. 5. CONCLUSION We present here database schema integration to realize interoperable GIS application with capability to provide a uniform access to multiple heterogeneous information sources. Thus database schema integration is the major activity to accomplish simplifying relocates geospatial data regardless of platform and format. The translation and transformation with XML technology makes it possible for cross platform sharing of information as known GIS software their own kind of database to hold spatial data, thus created in different format and schema of database. Accordingly semantic heterogeneous are on the topics of database schema integration concern. 6. REFERENCES [1] Jones, M. and Taylor, G. Metadata Spatial data handling and integration issues, School of Computing, University of Glamorgan, Technical Report Series, CS-03-01, (2003). [2] Cruz, I.R.; Huiyong Xiao; Feihong Hsu. An ontology-based framework for XML tic Integration. Proceedings of International Database Engineering and Applications Symposium, IDEAS '04, (7-9 July 2004), [3] Suryana, N. and Norayu Abd Ghani. Role of Clearinghouse Server Infrastructure for GIS Interoperability. Business Information System. GI Edition, Lecture Note in Informatics, (June 2006). [4] [5] [6] [7] Milan Trnini. OO, XML, and GML: Are angle brackets a flexible modeling material? Using XML comfortably among objects. JavaWorld.com, Internet source: xml.html. 09/05/05. [8] Regina Motz, Problems in the Maintenance of a Federated Database. Computer Science Society, SCCC Proceedings. 22nd International Conference of the Chilean on Nov, , [9] Mingxia Gao,Chunnian Liu, and Furong Chen. An Ontology Search Engine Based on Semantic Analysis. Information Technology and Applications, (Third International Conference, ICITA 2005 July 4-7, 2005, vol: 1, [10] Dario Bonino, Fulvio Corno, Laura Farinetti, and Alessio Bosca. Ontology Driven Semantic Search. WSEAS Transaction on Information Science and Application, Issue 6, vol: 1, (December 2004), [11] Chi-Chun Pan, Prasenjit Mitra, Peng Liu. Semantic Access Control for Information Interoperation. Proceedings of the eleventh ACM symposium on Access Control Models and Technologies. (2006), [12] Agustin Buccella, Alejandra Cechich, and Nieves R Brisaboa, An ontology Approach to Data Integration (Oct 2003). [13] M.Andrea Rodriguez and Max J.Egenhofer, Determining Semantic Similarity among Entity Classes from Different Ontologies. IEEE Transactions on Knowledge and data Engineering, vol 15, no 2, March/April