Version 1.0 Date 20032006 Status Draft Confidentiality

SOCRATES PROGRAMMEMINERVA : (ODL and ICT in EDUCATION)CABLE: Case-based e-learning for educators (109883 - CP -1-2003-1- IT - MINERVA – M)D.13 – Technological ReportVersion number 1.020/03/2006Executive summaryName Technological ReportVersion 1.0Date 20/03/2006Status DraftConfidentialityParticipant Partner(s) PdT, UHI, SydvastAuthor(s) Dario BoninoWork Package ?Distribution List All partnersAbstract This report provides a description of the technological solutions adopted by the CABLE project for deploying the learning services and methodologies designed during the project. Such solutions com-pose the so-called CABLE e-Learning framework which includes a VLE, a semantic correlation module and a case study repository. This report analyzes in some detail the design choices lying under the framework design and in particular shows how the involved modules have been implemented as well as how the same modules interact with each other to accomplish their tasks. The report is or-ganized as follows: in section 2 a brief introduction is provided de-scribing the report organization, the CABLE basic principles and the expected functionalities of the CABLE framework. Section 3 intro-duces the logical architecture of the framework while section 4 de-scribes the interactions between the different modules, referring the same interactions to real world usage scenarios. Finally sections from 5 to 7 describe in more deep detail the modules involved, also providing a description of involved technologies and related advan-tages.Keywords VLE, integration, development, semantic metadata, semantic correlation, interaction, usage scenarios, deploymentPrevious Versions -Version Notes 1.2Table of contentsTable of contents (3)Acronyms and Abbreviations (4)1 Abstract (5)2 Introduction (5)Architecture (6)3 System4 Modules interactions and usage scenarios (7)5 The semantic module (mH-DOSE) (9)6 The Bodington e-Learning system (11)7 The CSDB repository (12)8 References (13)Acronyms and AbbreviationsAcronym Description VLE Virtual Learning Environment: a general term for a managed access, web based learning platform which provides support to learning materials andsupport for the student and tutor during the learning process.CSDB Case Studies Database: the DB which houses the assembled CABLE case studiesURL Universal Resource Locator: the unique networked address of a resource on the WebTomcat The servlet container that is used in the official Reference Implementation for the Java Servlet and Java Server Pages technologies, developed by Sununder the Java Community Process (APL)./tomcatJ2EE The Java 2 Platform, Enterprise Edition, which includes web services sup-port and an SDK. /j2eePGSQL A sophisticated relational database management system, OS, which runs on a variety of platforms. Bodington An OS (Mozilla type) VLE platform built upon J2EE and PGSQL/ORACLE. ORACLE A powerful, object-oriented relational enterprise database management system, commercial, which runs on a variety of platforms:H-DOSE Holistic Distributed Open Semantic Platform, an open source project pur-sued by the e-Lite research group of the Turin’s Polytechnic.mH-DOSE Minimal H-DOSE, the H-DOSE reworking developed in the CABLE pro-ject.UML The Universal Modeling LanguageICT Information Communication TechnologyUHI University of Highlands and Islands (Scotland)A Relational Database Management SystemRDBMS /DBMSSourceForge A widely known open source software repository.1 AbstractThis report provides a description of the technological solutions adopted by the CABLE pro-ject for deploying the learning services and methodologies designed during the project. Such solutions compose the so-called CABLE e-Learning framework which includes a VLE, a se-mantic correlation module and a case study repository. This report analyzes in some detail the design choices lying under the framework design and in particular shows how the involved modules have been implemented as well as how the same modules interact with each other to accomplish their tasks. The report is organized as follows: in section 2 a brief introduction is provided describing the report organization, the CABLE basic principles and the expected functionalities of the CABLE framework. Section 3 introduces the logical architecture of the framework while section 4 describes the interactions between the different modules, referring the same interactions to real world usage scenarios. Finally the sections from 5 to 7 describe in more deep detail the modules involved, also providing a description of involved technolo-gies and related advantages.2 IntroductionThis report aims at describing in sufficiently deep detail the technological components devel-oped in the context of the CABLE project. These elements constitute the basis upon which the CABLE intelligent tutoring system is built, providing a tangible means that allows the analy-sis of experientially based case studies and the use of these case studies and other pedagogic and didactic data into a shared, Europe-wide, e-Learning framework.This report presents the technical issues involved in the project, describing the general archi-tecture of the e-Learning system, and then analyzing the several components involved. For each of them a functional description is provided, accompanied by a detailed study of the in-teractions with other modules, adopting well known UML formalisms such as use cases and interaction diagrams.The CABLE e-Learning system is built around two core entities, namely the case studies and the didactical modules. Users of the system can be students, authors of didactical modules and contributors of new case studies. The experience and implicit knowledge contained in the case studies, and the explicit knowledge contained in the didactic modules, need both to be han-dled by the system, in an intelligent manner, thus allowing to discover new relationships, shared concepts, methodologies, learning paths, etc. As a consequence, case studies and di-dactical modules are categorized by formal metadata, resorting to the same, domain specific, ontology or semantic network conceptualization.Classification of learning objects is a dynamic process, influenced by interactions and feed-back with users. The on-line courses shall therefore adapt themselves, automatically, to new case studies, and emerging common practices. To respond to this need of dynamism in col-lecting, correlating and publishing learning material, the technological platform shall be flexi-ble and intelligent enough, and shall be able to automatically discover new knowledge, as well as to autonomously link resources, according to their semantic classification. From these requirements emerge the architectural organization and the technological solutions described in the following sections.3 System ArchitectureCABLE is both a project and a technical framework for supporting the learning methodolo-gies experimented and developed during the project execution. The framework has a well-defined ICT (Information Communication Technology) infrastructure, which reuses, as much as possible, already available, effective solutions, avoiding to reinvent the wheel.Three main components are defined, composing the basic structure of the CABLE platform architecture: an e-Learning environment, a repository of case studies and experience-based learning data, and a semantic module able to leverage formal metadata associated to both learning objects and case studies, for composing and discovering associations between courses and practice examples.As the domain of application for the CABLE framework requires growing experiences of us-ers and teachers by allowing comparison and sharing of similar case studies and solutions, a semantic module has the responsibility of automatically establishing correspondences be-tween new learning paths and existing case studies, as well as the ability to correlate, at run-time, newly added case studies to the already existing cases and learning modules. Figure 1 shows the logical organization of the CABLE framework.Figure 1. The logical architecture of the CABLE framework.The VLE (Virtual Learning Environment) module is the Bodington learning environment [1], a cutting-edge, open source, e-Learning system widely adopted by UK universities (e.g., UHI, etc.), it is entirely developed in Java and runs on top of the Apache Tomcat servlet container [2].The CSDB (Case Studies Repository), instead, is a Java web application developed from scratch during the project execution.Finally, the semantic module mH-DOSE is implemented by a customized, minimal version of a publicly available, open source, semantic elaboration platform named H-DOSE [3,4].4 Modules interactions and usage scenariosThe basic interaction flow is designed to be two folded, i.e., to support two different informa-tion needs: finding case studies from a well defined learning resource in the VLE or finding case studies relevant with respect to a significant example (search for related case studies). In Figure 1 these two operational paradigms correspond, respectively, to the left most and to the right most user interfaces. As can easily be noticed both processes are mediated by the seman-tic module, which lies in the middle.The two paradigms can also be described in form of UML interaction diagrams. In the first case (Figure 2), a user, or a teacher, uses the VLE to participate in some learning activity. At a time, in the e-Learning process, case studies shall be analyzed for better understanding how to tackle a given pedagogical scenario. As the CABLE framework hosts many case studies pro-vided by several organizations, in a Europe-wide environment, the resources relevant to the learning module are extracted from the cases repository at runtime. Among other advantages, this allows to automatically take into account newly added knowledge, in a transparent way. Therefore, following the user request, the VLE contacts the semantic module for finding rele-vant case studies, providing, at the same time, a conceptual description of the learning object currently accessed by the user. The semantic module retrieves from the case studies repository all descriptions of cases studies that match, at least partially, the VLE specification. Then, by applying ontology navigation techniques, it ranks the retrieved results and provides back to the VLE a list of URLs, pointing at relevant examples. The VLE retrieves the case studies and presents them to the user, in a proper setting.Figure 2. The "VLE to case studies" interaction diagram.In the second scenario (Figure 3), the user is already accessing the case studies repository, as an example for consulting a well known solution to a specific pedagogic issue. Once read how the solution had worked and what is the scenario in which the same solution has been applied, the user might want to find whether the just learned approach has been successfully applied to other, similar, situations. He/She selects the “find related case studies” button on the user in-terface for retrieving similar cases (Figure 4).The button pressing causes, inside the repository, the retrieval of the conceptual description of the case currently viewed. Such a description is passed to the semantic module which, in turn, extracts from the repository a set of candidate case studies, on the basis of the initial semantic specification, taking into account ontology relationships and concepts. Then, as in the former scenario, the semantic module ranks retrieved results and provides a list of relevant URLs to the repository application. The repository retrieves and organizes the corresponding case stud-ies and presents them to the user, as a result.As can easily be noticed, in both cases there are no predefined matches between case studies, or between case studies and learning objects. Instead, they are discovered at runtime, by com-paring the respective conceptual descriptions.Figure 3. The "case study to case studies" interaction diagram.As the comparison is ontology-driven, not explicit associations can easily be discovered thus leveraging the power of semantics for providing conceptually relevant results (which are hopefully more relevant than the ones that would be extracted by applying simple keyword matching techniques).Figure 4. The "Find related case studies" user interface.5 The semantic module (mH-DOSE)In order to implement the functionalities required by the semantic module of the CABLE framework, an open semantic platform [3,4], developed by the e-Lite research group at the Politecnico di Torino (PdT) and available as an open source project at SourceForge [5], has been strongly customized, removing all the superfluous functionalities and modules. The ini-tial platform (depicted in Figure 5) includes, in fact several modules related to the storage of semantic descriptions (done by the case studies repository, in CABLE) and to the automatic indexing process, that in the CABLE framework is replaced by the creation of learning ob-jects and cases descriptions done by pedagogic actors/users.Figure 5. The original H-DOSE platformThe resulting platform is a light-weight version of H-DOSE (mH-DOSE) and is able to sup-port scalability and performances as required by the whole framework (Figure 6). In particular only two modules are actually required to ensure semantic correlation functionalities: the Search Engine service and the Expander service. The former offers the functionalities re-quired for managing both the “VLE to case studies” scenario and the “case study to case stud-ies” scenario, while the latter provides support for ontology navigation and automatic seman-tic correlation of involved entities.The Search Service implements the conceptual search functionalities required by the two automatic correlation paradigms deployed in CABLE. In both cases the process starts with the specification of a conceptual description. Such a description is firstly “expanded” through ontology navigation (i.e. using the Expander) and then used to retrieve relevant case studies stored in the CSDB. The CSDB, usually provides, as a response, a set of case studies that have been tagged with metadata similar to the description initially provided by the Search Engine. However such case studies still have to be ranked by the Search Engine in order to present most relevant resources in the first positions. The ontology-based expansion as well as the ranking performed by the Search Engine justifies why the semantic module of the plat-form is always involved in finding experience-based cases, even when already starting from a case study.The Expander service implements, inside the CABLE platform, the ontology navigation op-erator. It basically accepts as input a simple conceptual description and performs the naviga-tion using the CABLE ontology. The result is newly a conceptual description, which takes into account the not explicit knowledge encoded in the ontology. The expansion operator can be proficiently applied in the search phase, whatever being the interaction paradigm used. When a user specifies a query, by directly selecting concepts, in fact, the resulting specifica-tion is usually composed of few concepts. In this scenario the expander can browse the ontol-ogy, according to semantic relationships, and expand the query specification by adding rele-vant, related concepts. This operation potentially allows the system to retrieve resources which are interesting for the user but that, without expansion, would not have been retrieved because they do not have direct associations with the concepts originally composing the query.Figure 6. The CABLE semantic module.The semantic module of CABLE (mH-DOSE) has been entirely developed in Java and is de-ployed as a set of interacting Web Services running on an Apache Tomcat servlet container, by means of the Apache Axis framework [6]. The module is quite efficient being able to per-form an ontology navigation in few milliseconds (30 ms in mean time), thus respecting the scalability requirements stemming from the framework design.6 The Bodington e-Learning systemThe Bodington e-Learning system (Figure 7) is an open source VLE system initially devel-oped by the University of Leeds and now adopted by many UK universities as the Leeds Uni-versity, the Oxford University and the UHI Millennium institute. It has been chosen among several alternatives mainly because it is widely scalable, freely available, well tested and at least one project partner has a considerable skill in offering distance learning services by us-ing it.The complete discussion of motivations for choosing Bodington as VLE are out of the scope of this document, instead, the report is focused on how this VLE has been integrated into the CABLE framework.Figure 7. The Bodington e-Learning system.In CABLE Bodington offers two different functionalities: the fruition and publication of learning modules and the semantic classification of them. The first functionality does not re-quire any modifications to the standard e-Learning platform, since its main design goal is to provide learning materials to students. The second part, instead, requires the development of an additional publication menu available to teachers, which enables the semantic classifica-tion of newly added objects. Basically for each new resource a teacher, or better, a user hav-ing publication rights, may add a semantic description by choosing some concepts defined in the CABLE ontology. These concepts are listed into a simple directory (a list-box), that makes the selection fairly easy.Whenever some content has been so tagged, the VLE exploits a newly developed module for offering a “search for related case studies” menu, which provides access to the knowledge base of pedagogic experiences gathered by the CABLE partners and stored in the CSDB re-pository.The Bodington VLE is a J2EE application running on Apache Tomcat and using the Post-greSQL DBMS [7] as a persistence backbone. This peculiarity makes particularly easy and efficient the integration of the learning system into the CABLE framework as also the other modules are J2EE services deployed on Tomcat7 The CSDB repositoryThe CSDB repository is mainly composed of a RDBMS for storing case studies and related metadata and of a web application UI for providing an easy-to-use access to the DB reposi-tory (Figure 8). As the flexibility and security constraints defined in the design phase of the project already identified a UNIX solution, and given that Bodington is based on the Post-greSQL RDBMS which respects all the technological requisites for the CSDB, the case study repository has been designed and developed using the same storage system. In addition, being the J2EE environment one of the most diffused web application frameworks applied world wide, the UI web application has been developed as a Java Web Service, running on Tomcat and Axis.This design choice allows, among other advantages, to deploy all the CABLE framework modules on the same technological infrastructure, which is completely Open Source and freely available. This allows an easy deployment of the framework in several learning envi-ronments, an constitutes one of the technological strengths that allow the CABLE project to be sustainable even when the European Commission funding will cease.Figure 8. The CSDB web interface.8 References[1] The Bodington Virtual Learning Environment, [2] The Apache Tomcat servlet container, /tomcat[3] The H-DOSE platform, [4] DOSE: a Distributed Open Semantic Elaboration Platform. D. Bonino, F. Corno, L. Fari-netti. ICTAI 2003, The 15th IEEE International Conference on Tools with Artificial Intelli-gence, November 3-5, 2003, Sacramento, California.[5] The SourceForge open repository, [6] The Apache Axis Web Services framework, /axis/[7] the PostgreSQL DBMS, 。