XMLEase, a Speed-Efficient Approach to Courseware Data Representation In Relational Databases

 XMLEase, a Speed-Efficient Approach to Courseware Data Representation in Relational Databases Atilla Elçi and Behnam Rahnama Department of Computer Engineering and Internet Technologies Research Center; Eastern Mediterranean University; Gazimagusa, T.R.N.C {atilla.elci; behnam.rahnama}@emu.edu.tr Abstract. Courseware content consists of hierarchically ordered documents which may include arbitrary-size text, image, video, audio, etc. The structure of pre-requisite courses, inclusion of relevant information in the table of contents, connecting through hyperlinks to items and access to alternative sections for further study are several sources of hierarchies in a courseware. Using relational models for keeping the data of such hierarchical documents is inefficient due to the limitations of relational databases. Object oriented databases ease storage of objects but lack fast access methods. In such cases a better technique is required to speedup data representation and retrieval within relational databases. XMLEase lets the data node to be kept with its successors up to a defined level of hierarchy. Each node includes links to the document objects kept in a normalized table. This method has been implemented and evaluated in comparison to similar methods as a successful representation method for XML data within relational databases. Nevertheless it suffers of a little bit extra space usage for n successors’ links, but it shows a fair linear performance independent to the depth of hierarchy with great data retrieval efficiency. 1 Introduction The use of XML in courseware has become common place. This is so to the extent that new courseware is being created employing XML, recent learning management systems (LMS) are designed to ride on XML, and earlier LMS are being retrofitted to take advantage of XML way of data marking. Even the newer courses instructing authors on how-to of courseware development cover XML file system prominently. Let’s take a look at some samples. Courseware initiative EduMine offers its “Courseware Development for Authors” course which is designed as a user guide and reference for authors and developers of EduMine courses. Topics covered by Courseware Development for Authors include ‘XML tag references for courseware documents’ as one of the main subjects to muster among mainline topics such as introduction to courseware development, courseware preparation and structure, and courseware content development [24]. Similarly, Myrnham Associates’ Courseware Authoring Suite consists of two applications, CourseComposer & CourseManager, both of which utilize the XML Docbooks DTD (of standards body OASIS, [25]). Among many features of CourseComposer, XML capability plays a central part. The files CourseComposer produces can be printed directly using any of the Open XML tools freely available on-line using a number of different openly available XSL Style sheets. Alternatively, CourseManager may be used to supply a MS Word document. CourseComposer enables its user to leverage XML-based industry standards (CVS, WebDAV, XML, and DocBook) to develop a library of content [26]. On the other hand, legacy LMS owners have been migrating to XML technology. For example, Eastern Mediterranean University’s LMS was originally based on HTML design however has since switched to XML with its second version [27]. The University of Southern Queensland (USQ) switched to XML based GOOD (Generic Online/Offline Delivery) system to availing cross-media publishing, automated publishing process, on-demand outputs, and computer managed assessment [30].  Likewise, numerous recent LMS were designed from day one to employ XML prominently in their courseware [28]. References to several market leader courseware development environments may be reached through the reference [27]. Our point of view to factor out control aspects from IEEE’s Learning Technology Systems Architecture (LTSA) in order to render it pure XML-based can be reviewed in [29]. In the following, we first mention the benefits of XML use as perceived by practitioners. Then, the essence of XML technology as well as the efficiency concerns in maintaining XML data in databases is introduced. We then explain our new novel method for efficiently storing and accessing XML data in relational databases and show its virtues. 2 XML Initiative for Courseware The utility of XML base for delivery systems and courseware itself appears to be undeniably extensive. Through XML, LMS is able to provide services all around. For example, the USQ’s GOOD System enabled cross-media publishing from a single document source stored in a content management repository. It’s reported that the GOOD system has enabled USQ “to replace its resource-intensive legacy production system for courseware with a single document source system based on the XML (eXtensible Markup Language) standard.” Academic content specialists are able to edit their XML documents, and to generate HTML, PDF and PostScript outputs on demand. Similarly, taking advantage of the XML basing, GOOD system is also used for other applications across the University, including the cross-media publication of the Handbook, course information documents, course specifications, past examinations and Computer Managed Assessment (CMA). CMA employs single-source XML tests in generating all deliverables (PostScript, PDF, HTML and interactive CDs) including student feedback. The metadata required for managing all aspects of assessment delivery and marking is also provided as XML[30]. On the other hand, XML-tagged courseware documents are structured within uniform, comprehensive parameters with the substantive content able to be treated discretely from its layout and presentation.  The document layout is generated by applying XSLT (eXtensible Style Sheet Language: Transformations) to the XML-tagged content. XSLT is a language for transforming XML documents into their target formats (for example HTML for Web delivery). The GOOD cross-media production system uses XSLT to simplify and fully automate the task of publishing content in multiple formats. The GOOD “rendering engine” automatically converts XML content into PostScript and PDF for print delivery, and into HTML for online and CD delivery. The GOOD source documents are edited using a highly structured XML editor [30]. Setting out to develop an XML-based database courseware, the authors realize at the end that indeed they developed a framework for future courseware [31]. The presented framework allows users to choose their own different educational path from the one that can be obtained on the module dependency graph, decreasing this way the time needed to understand a certain topic can be much lowered that in a linear educational path. The possibility to adapt the content of the course to the knowledge level of a student assure that each student will get a personalized content, containing only the topics that he has to study, without obligating him to “scan” through the content in order to identify those topics. Another benefit of the presented framework is the possibility to adapt the presentation form to user preferences as well as to different specific restrictions (hardware, connection speed, user disabilities etc.), providing a content that is easier to study. The educational value of courses implemented by using this framework increases considerably, due to the presented facilities. Web-based courseware generation meanwhile has become increasingly complicated as this also introduces additional challenges into the courseware authoring and publishing process. There have been institutional efforts to jointly develop platforms in order to meet this challenge. For example, [32] presents a model platform for facilitating the courseware generating process taking advantage of recent Internet protocols and industry standards. The study adopted WebDAV, a collaboration-friendly Internet protocol, to support collaborative courseware authoring, XML to represent meta-data of course contents, and JSP to effect dynamic courseware presentation. In the line of developing an XML-based platform, the Sharable Content Object Reference Model (SCORM) initiative has a prominent part. SCORM is a set of interrelated technical specifications built upon the work of the AICC, IMS and IEEE to create one unified 'content model'. These specifications enable the reuse of Web-based learning content across multiple environments and products [33]. XML-based specifications supporting learning technologies have been developed by the SCORM initiative and distributed through the Advanced Distributed Learning Initiative (http://www.adlnet.org/) Network. Chapter 5 of the current SCORM specification (http://www.adlnet.org/Scorm/docs/SCORM_2.pdf) presents the Course Structure Format. CSF is "an Extensible Markup Language (XML)-based representation of a course structure that can be used to define all of the course elements, structure, and external references necessary to move a course from one LMS environment to another..." Performance of XML-based platforms then has become important for the success of LMS and courseware implemented. This paper takes up this performance issue and proposes a new model to remedy it. Rest of the paper is arranged as follows. Next section discusses an XML data model in terms of a small e-learning example. In the following section, performance issue of XML implementations is introduced in terms of storage and access efficiency. In section 5 we mention existing approaches for XML data representation in relational databases. Then on we introduce the proposed method, XMLEase, affording storage and access efficiency. In section 7, common queries and sample SQL syntax for processing of XML data through XMLEase model are given. Section 8 contains analytical and experimental results displaying that XMLEase is a brilliant choice for both access and space efficiency. And finally, we conclude the paper with findings. 3 XML Data File In this part we show a short XML data and its hierarchical representation. The XML data graph for the following XML document given in Figure 1. <?xml version="1.0" encoding="ISO-8859-1"?> <?xml-stylesheet type="text/css" href="css/docbook.css"?> <!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.3//EN" "file:///../content/rules/docbookx.dtd" [<!ENTITY % global.entities SYSTEM "global.xml">%global.entities;]> <book> <bookinfo> <title> XML and Databases </title> <keyword> XML, Mapping </keyword> <author> <firstname> Ronald </firstname> <surname> Bourret </surname> </author> <corpname> Ronald Bourret </corpname> <pubdate> 1999-2005 </pubdate> <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="Myrnham.legalnotice"/> </bookinfo> </book> Fig. . A small XML Document and its corresponding data graph In Figure 1, the root “customer” is regarded as the 0th level of the XML tree (we start counting from the root), the “name” is in the 1st level of hierarchy, etc. In the following section, performance issue of XML implementations will be introduced in terms of storage and access efficiency. 4 Performance of XML Technology Implementations Extensible Markup Language (XML) is a simple, very flexible text format derived from SGML [1]. XML data is of a hierarchical data structure, hence the query languages such as XQuery, YATL, XML-QL and XML-GL were proposed to easily write and parse the XML data [2, 3]. In order to facilitate the task of querying XML documents, efficient storage models for storing XML data in relational databases some methods such as Edge, Monet, XRel and XParent were proposed [4]. A comparison studied by Jiang et al. shows that the XParent is the best approach in terms of storage efficiency while its access efficiency is poor [4, 16]. XParent method scatters XML data into three tables, namely, “Text”, “Element”, and “Path”. In that case, to access the XML data efficiently one needs to “join” these three tables. Several allies of approach have evolved to enhance efficiency of handling XML data. DBXML [12], de-normalized relational tables [22], and support of XML schema within SQL codes [18] are for processing XML data in relational tables. In addition, simplified query based techniques such as RelaXML [13] and XML data scheme [14, 15] could also help maintain XML data. Moreover, processing methods for indexing XML data [17, 19] in order to raise the query efficiency such as search queries [20] are other area of interest. However, few methods could be applied for both access and space efficiency models [16, 23]. Some mapping schemes [11, 14, and 21] try to meet both requirements; nevertheless the complexity of the models is highly increased. In this paper, we examine a database schema called XMLEase as a novel approach to XML data representation model. This approach regards an XML file as consisting of terms of hierarchical data structure, each node of which is made up of “Label”, “Data”, and ancestor link fields. Each node is then converted to a record in a relational table. As a node is stored in a table together with its redundant pointers to its ancestors in the XML tree, the “join problem” is effectively removed. This causes some increase in space usage while affording higher access efficiency. This approach benefits from avoiding any requirement for data field processing of relational data as it’s the case with XParent method, thus allows running a single SQL query to store or retrieve XML data. Existing approaches to XML data representation in relational databases are taken up in the following section. XMLEase is further introduced in Section 6. 5 The Existing Approaches There are many models dealing with hierarchical data structures that may be utilized in representation of XML data in relational DB tables. Among more common ones, the “Nested Sets Model” [7, 8] assigns sequential numbers to nodes for two indices to effect bidirectional search queries. In the “Materialized Path Model” [5, 6], the “address” of a node is stored within it, starting from the root, in the form of a partially-concatenated string of integers. The mapping between XML and relational data are treated in [2, 10], and [14]. Structure-mapping based XML Data Mediator [9] methods also belong in this category. Edge, Monet, XRel, RelaXML, and XParent are five prominent approaches to store XML data in relational tables [3]. The access efficiency of Edge is better than that of other methods, because it employs a single normalized table. On the other hand, XParent method based on Materialized Path Model shows the best space efficiency. XParent uses numerical tables in connection with the XML data table in order to avoid text processing on the string data kept in the materialized path table [4, 23]. 6 A Novel Method: XMLEase XMLEase is a novel efficient method that introduces a way of converting an XML tree to RDBMS tables. To understand this method better, let’s introduce redundant edges into the tree, so that each node is connected additionally to each one of its ancestors by a direct link, instead of just to its parent. This makes it possible to fetch an XML tag data of a sub-class for a specific property (such as if we need all the name spaces in the third level of XML tree) using a single query, without needing a join operation. Table 1 below displays the XMLEase model representation of the XML data of Fig.1 above. Table . XML data graph representation using XMLEase method (Identifier) (P1) (P2) (P3) (P4) book NULL NULL NULL NULL bookinfo book NULL NULL NULL title bookinfo book NULL NULL keyword bookinfo book NULL NULL author bookinfo book NULL NULL corpname bookinfo book NULL NULL pubdate bookinfo book NULL NULL include bookinfo book NULL NULL &XML and Databases title bookinfo book NULL &XML, Mapping keyword bookinfo book NULL firstname author bookinfo book NULL surname author bookinfo book NULL &Ronald Bourret corpname bookinfo book NULL &1999-2005 pubdate bookinfo book NULL . . . include bookinfo book NULL &Ronald firstname author bookinfo book &Bourret surname author bookinfo book The “Identifier” column contains a name for intermediate nodes and the immediate data in the case of leaves of the XML tree in the example. Other columns maintain pointers to ancestors, such as, P1 is the pointer to immediate parent, P2 is the link to the grand parent, and so on. For example, the record for “Name” has “book” as P1, and “NULL” for P2, P3 & P4. In this case, the depth of hierarchy supported by a relational database is limited by the number of parent columns. On the other hand, the bigger the number of ancestor columns, depending on the specific XML document, there will be more chances for “NULL” pointers, thus wasting storage. It is up to the user to pre-determine the level of hierarchy in order to keep efficiency high vis-à-vis the nature of the XML files in consideration. The user may reset the level at any time. Note that the scope of hierarchy in XMLEase is limited to the supported number of parents (number of pointer columns; let’s call it “a layer”) rather than keeping the complete depth of all the nodes. In order to fetch the nodes at deeper levels than the supported level, the scheme is applied to the successive layers in turn and query results are joined. As far as the model requires a single run for each layer of hierarchy, it is also access (i.e., time) efficient. This is taken up in the next section. It is conceivable that XML data hardly ever has a high number of hierarchies. Choosing the supported level as, say 10, would require just a single layer, thus generally utilized without requiring joins or re-runs. 7 Access Efficiency of XMLEase in Query Processing XMLEase speeds up query processing for insertion, update, retrieval and deletion of data. This is due to its keeping parent pointers with entities. The sample queries given below may help show this result. The examples in this section refer to the relational table given above. 7.1 Data Insertion In order to insert a new XML tag, say “Email”, as another property of “author”, one can easily run the following query: INSERT INTO xmltable VALUES(’Email’,’author’,’bookinfo’,’book’,0); The entity <Email, author, bookinfo, book, NULL> will be inserted. Since the ancestors’ addresses are kept in the pointers of this entity, one does not need to update the parent nodes while inserting a node into an XML tree. Consequently, a parent’s field need not be updated in order to add a successor node. Thus, insertion is completed in single-run query. 7.2 Data Update To update a data value in the relational table, one needs only to run a simple single-run query. For instance, to change the corpname value to “&E-Trainers, Ltd.”: UPDATE xmltable SET xmltable.Identifier = ’&E-Trainers, Ltd.’ WHERE xmltable.P1=’corpname’ and xmltable.P2=’book’; 7.3 Data Deletion In order to delete a tag, one can simply delete its specific record. Since the ancestors’ addresses are kept in the pointers, while deleting a node from an XML tree its ancestors’ nodes need not be updated. However, a node and its sub-tree can be removed with single-run query without requiring looping or recursion. This is done simply by inner select deletion if DBMS supports it, or by deletion of entities explicitly selected through parent pointer being the deleted tag. 7.4 Data Retrieval With the XMLEase table representation, the following two major queries need be covered: a) ‘Fetching all the children of a node, say, “bookinfo” that are 2 levels down’ can be done with the following query: SELECT Identifier FROM xmltable WHERE P2 = ’bookinfo’; b) To find all of the leaves of the sub-tree of “book”, the following single query can be run: SELECT Identifier FROM xmltable WHERE P1 = "book" or P2 = "book" or P3 = "book" or P4 = "book"; As seen above, in both cases only a single-run query is required. The examples given above help elucidate the access efficiency of XMLEase. In the following section, space efficiency of XMLEase will be shown; analytical formulation and experimental comparison against other similar methods will be displayed. 8 Space Efficiency of XMLEase 8.1 Analytical Results Among all existing methods, the materialized path is superior in terms of access efficiency and space used to keep XML data. Therefore, the XMLEase method is compared against the materialized path. In doing so, we have derived analytical formulation of space usage of both methods in a recent study [23]. Here we summarize the results. Let’s denote the space used by XMLEase method without indices by; the total number of nodes in an XML tree by N; the level of hierarchy by L; and the size of the record identifier by S. Then: () Taking into account the indices on the table, space used by XMLEase method,, is expressible as follows, where Sp is size of pointer, n is the maximum possible number of nodes at each level, is equal to the size of indices on each column for level parent (non key attribute), k is the supported level of hierarchy (i.e., number of parent columns,), and after a simplifying assumption of k = L: (2) Whereas, the space usage for the materialized path without indices is as follows if Ps is size of each path entity and S is the size of each dot-separated integer entity: (3) On the other hand, considering indices in the table, and after a simplifying assumption on the size of each entity,is given by the following: (4) The comparison of formulas () & (3) for space usage without considering indices is straightforward: XMLEase is superior to the materialized path method. In comparing the cases with indices of XMLEase (2) and the materialized path (4), we compute their difference as follows: (5) It can be shown thatis always positive [23]. Thus, XMLEase method uses less space than the materialized path method with or without taking indices in to consideration. 8.2 Experimental Results In the experiment reported below, space usages for different methods are being compared. As the first test case, let us consider database tables without taking indices into consideration. A simple program was written for working with MySQL DB in order to fill the relational tables by the same XML data. The chart given below displays the space used versus levels of hierarchy by different methods without considering indices. Fig. . Space used (size) versus levels of hierarchy by different methods without considering indices It is clear that the experimental values follow the analytical estimations: XMLEase method uses less space than materialized path and redundant arcs methods. However, it consumes more space than naive, adjacency list and nested sets methods but these are not as access efficient as XMLEase. The chart given below displays the space used versus levels of hierarchy by different methods if indices are taken into consideration. Fig. . Space used (size) versus levels of hierarchy by different methods considering indices In the cases where indices on the tables are taken into consideration, the experimental results show that XMLEase is better than all methods except the naïve one in agreement with analytical estimations. However, naïve method is access inefficient as already indicated. 9 Conclusion There would be several alleys open to managing semi-structured data as XML-based e-learning material in a content management repository such as using relational, object and native XML databases. Of these, the relational database is the one that has a solid scientific base, the most settled technology and the most commonly used. Representing XML data in terms of the rows and columns of a relational database table is inherently quite inefficient. Similarly, object-oriented databases cater for storage of objects but lack fast access. In these cases, a better technique is required to speedup representation and retrieval of hierarchical data in relational databases. This paper presented XMLEase, a new access- and store-efficient technique to maintain XML data in relational databases. Various models of representing XML data in relational databases are considered; their performances are compared against that of XMLEase. The best of the methods using existing models such as Nested Sets and the Materialized Path Models is the XParent. Yet, the XParent is known to suffer from low performance for path interpretation. The XMLEase method introduces redundant links between a node and its ancestors so that the children of a given “Label” at any level below the node in the XML data file can be retrieved efficiently. The XMLEase places these links as pointers in separate columns. In this approach, the XML data tags refer to their ancestors instead of to their successors as it is in the case of XParent method. It is shown that the fundamental DB data manipulation operations such as insert, update, delete, and retrieve are carried out efficiently with a single-run query using XMLEase. Additionally, deleting a node does not require updating the previously stored parents. Space efficiency wise, the small extra space used for pointers appears to be the only drawback of this approach. Yet, wasted storage in XMLEase use is much less than that of the other methods used for dealing with XML data. Furthermore, it is clear that the XML data servers value ‘access efficiency’ and speed more than they would mind using a little extra storage space. Experimental results follow the analytical expectations. XMLEase method is more efficient than redundant arcs and materialized path methods in terms of space usage. It performs better than all others in the case of access efficiency. References W3C Organization, Extensible Markup Language, http://www.w3.org/XML/ and http://www.xml.org/. visited: 10.03.2006. Katz, H., Chamberlin, D., Draper, D., XQuery from the Experts: A Guide to the W3C XML Query Language, Addison-Wesley Professional, 1st edition (2003). Bonifati, A., Ceri, S., Comparative analysis of five XML query languages, vol. 29(1), SIGMOD Record, (2000) 68–79. Jiang, H., Lu, H., Wang, W., Xu Yu, J., Path materialization revisited: an efficient storage model for XML data, Proceedings of the Thirteenth Australasian Conference on Database Technologies, Australian Computer Society, Inc., Melbourne, Victoria, Australia Celko, J., Trees in SQL, http://www.intelligententerprise.com/001020/celko.jhtml?_requestid=123193. visited: 10.03.2006. Tropashko, V., Nested Sets and Materialized Path, http://www.dbazine.com/tropashko4.shtml. Celko, J., Trees and Hierarchies in SQL for Smarties, Morgan Kaufmann (2004). Furnas, G., Zacks, J., Multitrees: Enriching and Reusing Hierarchical Structure, Human Factors in Computing Systems, Boston, Massachusetts USA, (1994), 330–336. Bohannon, P., Freire, J., Roy, P., Simeon, J., From XML schema to relations: a cost-based approach to XML storage, Proceedings of the 18th International Conference on Data Engineering, IEEE Proceedings (2002) 64–75. Chaudhuri, S., Chen, Z., Shim, K., Wu, Y., Storing XML (with XSD) in SQL databases: interplay of logical and physical designs, IEEE Transactions on Knowledge and Data Engineering, vol. 17(12), (2005), 1595–1609. Maghamez, A., Hu, G., Multi-resolution indexing for XML data, 3rd ACIS International Conference on Software Engineering Research, Management and Applications, (2005), 206–211. Tzvetkov, V., Xiong W., DBXML - Connecting XML with Relational Databases, CIT 2005, Proceedings of the 5th International Conference on Computer and Information Technology, (2005), 130–135. Knudsen, S.U., Pedersen, T.B., Thomsen, C., Torp, K., RelaXML: bidirectional transfer between relational and XML data, Proceedings of the IDEAS 2005, 9th International Database Engineering and Application Symposium, (2005), 151–162. Fujimoto, K., Dao Dinh Kha, Yoshikawa, M., Amagasa, T., A Mapping Scheme of XML Documents into Relational Databases using Schema-Based Path Identifiers, Proceedings of the WIRI '05, International Workshop on Web Information Retrieval and Integration, Proceedings (2005), 82–90. Xing, G., Guo, J., Wang, R., Managing XML documents using RDBMS, Proceedings of the SNPD/SAWN 2005, Sixth International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, and First ACIS International Workshop on Self-Assembling Wireless Networks, (2005), 186–191. Qin, J., Zhao, S., Yang, S., Dou, W., Efficient storing well-formed XML documents using RDBMS, Proceedings of the 2005 International Conference on Services Systems and Services Management, vol.2, Proceedings of ICSSSM '05 (2005), 1075–1080. Shanmugasundaram, J., Shekita, E., Kiernan, J., Krishnamurthy, R., Viglas, E., Naughton, J., Tatarinov, I., Special section on advanced XML data processing: A general technique for querying XML documents using a relational database system, ACM SIGMOD Record, vol.30(3), ACM Press (2001). Nicola, M., Linden, B., Industrial session: XML support in relational system: Native XML support in DB2 universal database, Proceedings of the 31st International Conference on Very Large Data Bases VLDB '05, VLDB Endowment (2005). Amer-Yahia, S., Du, F., Freire, J., XML processing: A comprehensive solution to the XML-to-relational mapping problem, Proceedings of the 6th Annual ACM International Workshop on Web Information and Data Management, ACM Press (2005). Tatarinov, I., Viglas, S., Beyer, K., Shanmugasundaram, J., Shekita, E., Zhang, C., Research sessions: XML I: Storing and querying ordered XML using a relational database system, Proceedings of the 2002 ACM SIGMOD international conference on Management of data SIGMOD '02, ACM Press (2002). Weigel, F., Schulz, K., Meuss, H., XML data management and web discovery: Exploiting native XML indexing techniques for XML retrieval in relational database systems, WIDM '05, Proceedings of the 7th annual ACM international workshop on Web information and data management, ACM Press (2005). Balmin, A., Papakonstantinou, Y., Storing and querying XML data using denormalized relational databases, The VLDB Journal, The International Journal on Very Large Data Bases, vol.14(1) Springer-Verlag New York, (2005). Rahnama, B., A New Method to Represent Hierarchical Data in Relational Database Systems, Master of Science in Computer Engineering, Eastern Mediterranean University (2005) Houlding, Simon, W., Courseware Development for Authors, InfoMine Professional Development Courseware, Version: August 07 2001. http://www.edumine.com/xutility/html/menu.asp?category=xcourse&course=Xauthor, visited: 15.03.2006. DocBook. http://www.oasis-open.org/docbook, visited: 12.03.2006. Courseware Authoring Suite, Version: 1.01 by Myrnham Associates, http://www.xml.com/pub/p/427, visited: 13.03.2006. Aybay, I., Abdulova, V., A Learning Management System for Online and Hybrid Distance Education:  EMU_LMS v2, INISTA - Int. Symp. Innovations in Intelligent Systems and Applications, pp. 203 -206, 15-18 June 2005, Istanbul, Turkey. EduTools Course Management Systems web page: http://www.edutools.info/course, visited: 14.03.2006. Elçi, A., A Metadata for E-Learning Coordination through Semantic Web Languages, in Proc. of IV. International Symposium on Educational Technology IETC 2004, 24-26 November 2004, Sakarya, Turkey; in Proceedings (İşman, A. and Dabaj, F., editors) by TOJET 2004, pp.: 27-32. Also in http://www.tojet.net/articles/433.htm. The University of Southern Queensland, The GOOD (Generic Online/Offline Delivery) System, http://www.usq.edu.au/dec/staff/good.htm. visited: 15.03.2006. Bota, F., Corno, F. and Farinetti, L., "XML Based database courseware," in Proc. ICDE 2001 - 20th world conference on open learning and distance education, Düsseldorf (Germany), 01-05 April 2001. Also, at http://www.cad.polito.it/pap/db/icde01b.pdf. Qu, C., Gamper, J. and Nejdl, W., A Collaborative Courseware Generating System Based on WebDAV, XML, and JSP. http://www.kbs.uni-hannover.de/Arbeiten/Publikationen/2001/qu_Icalt.pdf. visited: 15.03.2006. OASIS Cover Pages, Shareable Content Object Reference Model Initiative (SCORM), http://xml.coverpages.org/scorm.html. visited: 15.03.2006. &Bourret Ronald surname firstname . . . &1999-2005 &XML, Mapping &Ronald Bourret &XML and Databases author include pubdate keyword corpname title bookinfo book