Future Dimension Of Learning Object Repository

Published Date: 02 Nov 2017

Abstract—Web has become an invaluable information resource with information explosion and searching a time consuming process. E-learning is day by day evolving with the web and has attained milestones like Mlearning and pervasive learning. LMS manage, document, track, covers and delivers educational resource termed as Learning Object (LO). LO’s stored in Learning Object Repositories (LOR) with metadata or catalogue. LO’s digital entities that act as an aid in edifying a subject in a syllabus. It contains information; objective and teaching methodology bind together. LOR’s have high variability on its description and its existing searching modules are not enough to fulfilling the user information requirements. This paper discuss lacuna of LOR’s and describes need of semantics web architecture in Learning object repositories through Semantic Learning Object Repository (SLOR).

Keywords—Learning object; Learning object repository;Learning management system; Learning content management system; Semantic Learning object repository; XML; Ontologies.

Introduction

Learning Object Repository (LOR) stores searches and distributes learning object and their metadata. It authorizes users to share, manage, and reuse learning resources. In LOR, learning resources are kept in object format thereby increasing versatility, search ability and storage customization of learning objects. LOR provide search interface through which humans can interact and also facilitates query interface which is used by software agents.

Learning Object is an independent object for learning that ranges from smaller bit size to full fledged module, from anything to everything for learning in specific context.[1] Learning objects are self contained, interoperable, reusable. LO can be aggregated to structure a new LO and are generally tagged with metadata. Metadata is data about data which give the description about data. It can be viewed as a set of properties or attributes necessary to explain the learning object. It makes the searching or exploring of learning object easier, by providing reusability and interoperability to LO’s [2]. In practice most commonly used metadata standard are IEEE LOM and Dublin Core metadata standards. The Dublin Core contains 15 elements for use in any resource description but it does not include attributes describing pedagogical perspective of a document, where as IEEE LOM consider it. IEEE LOM contains nine categories of data elements to describe a learning object- general, lifecycle, meta-metadata, educational, technical, rights, relation, annotation, and classification. Whereas Dublin Core standard include two levels: Simple & Qualified. The elements are Title, Subject and its Description, Document Type, its Source, Document relation, coverage, creator, Publisher, Contributor, permissions or rights, date, format, Identifier and Language. The elements of qualified Dublin Core include three categories-Audience, Provenance, and Rights Holder, additionally Qualifiers that refine the semantics of the elements in a way that may be used in Semantic Web [3].

Learners can connect to a LOR to publish or retrieve requisite learning object. Instructor can interact with LOR through an LMS to create and manage learning objects for the course. Learner can also use an LMS to access learning object. Learner can download a Learning Object from LOR and use it.

Learning Management System and Learning Content Management System

Learning Management System (LMS) is an e-learning environment that handles learning resource publishing, administration and delivery. LMS allows an organization to enroll user for the learning, delivery of Learning objects or resources, tracking and reporting of learning. It ensure reusability by utilizes LO’s, ensure goal progress and determine knowledge gained by the learner.

Learning Content Management System (LCMS) is a multi-user platform where developer can create, reuse, manage, store and deliver content from a Learning Object Repository. The process of designing e-learning courses in LCMS is supported with query processing techniques that explore metadata and semantic relationships to aid in LOs searching [4].

Learning object repositories (LOR) can be defined as database or storage metadata and links to attach to the actual learning resource. It provides benefits to an organization as well as to an individual by storing learning resources or LOs collectively; hence sharing and reusing of learning content; saves human time and effort. Individuals can also share learning objects and can also create and allocate the deliverables. Learning objects quality should be measured and maintained though LMS.

The following are the key feature that makes LMS and LCMS distinct:

Table 1: LMS vs. LCMS

Key Functionality

LMS

LCMS

Key Users

Learners, Administrators, Instructor

Content Developer, Subject Matter Designer, Instructional Expert[11]

Key Services

Course Publishing & Enrolment

Author and SMD course creation

Organizing reusable content

Not available

Primary goal of the system

Integrating LMS and LCMS

LO’s major issues are LO’s management and its deployment in Learning environment. The approach of integrating the features of LMS, LCMS and LOR, is emphasized upon in this section and other existing architectures were discussed as:

In the first architecture, LMS stores pedagogical documents in a dedicated database without any metadata linked to it. These stored documents will become unusable due to the lack of metadata linked to them at the time of retrieval of required pedagogical documents. This architecture is just a black box which takes input as pedagogical documents and lacks communication; hence user cannot share and reuses pedagogical documents.

Figure1. Black box based LMS architecture

The second architecture uses an open system approach where pedagogical documents are stored independently and provide transparency in sharing and reuse of pedagogical documents or LO’s. LCMS store both document and linked metadata generally termed as LOR. Metadata categorizes the pedagogical documents and provides faster retrieval [5]. LOR is generally not included as a part of learning environment and LMS is unusable to LOR and are two complementary concepts so we require some interface to bridge the gap and provide efficient search mechanism.

Figure2. Open system based LCMS/ LOR architecture

Third architecture is a service oriented mediator architecture in which the LMS, LOR and LCMS are three independent modules and layers; allowing communication between LMS and LOR.

Figure3. Service oriented architecture

In the architecture shown above LMS is the heart of the learning environment which manage deliverable and deliver the same to the users and LOR integrate LO’s in the learning process and the middle layer; mediator associate LMS and LCMS.

Depending of the users privilege the middle layer allow user to query the LOR from the LMs to retrieve required LO. Mediator layer allows required LO’s can be downloaded on local machine and transfer these LO’s from LOR into the LMS.

Further modified mediator architecture was discussed which uses LO retrieval from heterogeneous LOR’s. A mediator component was developed which accepts user query and transform into learning object repository specific query and forward the same to repositories for retrieval. This system enable searching not only in the system itself but it search group of repositories. An ontology based query rewrite method was discussed.

These commands are implemented in the following three methods where each layer offer service to mediator: leveled queries, are sent by Learning agent to mediator, in response to leveled queries wrapped layer sends repository specific queries to Repository, then Repository returns a set of URL to Mediator and forwarded further to the Learning agent.

A repository always does not contain a powerful cache system; hence it is rational to integrate a cache system with the mediator. If a query is sent for the second time, the mediators checks it in these connected repositories, but rather get the results from cache which is a set of URIs. Inclusion of Cache in mediator system has reduced run-time complexity significantly. The results of subparts of queries are also cached.

Shortcomings of Learning Object Repository

Although the Learning objects repository yields considerable advantages over LMS and LCMS, but it carry significant disadvantages as:

LOR need complicated hardware, software and highly expert staff; hence increased cost.

Frequent updation or replacement introduced for LO’s, stored in LOR is very complex process and leads to high management cost. The updation process should be properly managed to ensure enrichment of the LO stored in the LOR rather than inconsistent LOR. The updation involves high cost to train users and system administrators to use and manage the process.

LO lacks formal schema or standardization for the metadata learning objects and its metadata. Existing LOR uses LO’s collaborate reviews by expert registered uses to guarantee quality of the learning object which are generally expressed in natural language will become sometime difficult for machine to understand and use. Collaborate reviews though got an added advantage for enhancement of efficiencies through shared skills, knowledge transfer but is costly and time consuming process.

Existing LO metadata standards contributes towards the structure of learning object but it does not discover the dynamic semantics or meanings associated with LO.

Existing metadata standards have evidenced that delegating LO metadata to contributors for making decision on how to annotate learning objects may result in fragmentary and incomplete metadata records which makes LO usable, for software agents in automated or semi-automated scenarios [7].

Current metadata standards are not machine oriented or Natural language processing based however metadata only gives the content description rather than context or semantic searching. So there arise needs of Semantic Learning Object Repository which can overcome all these conflicts.

Techniques used in Semantic Web

Semantic Web is not a separate Web but its an extension of the World Wide Web, where information is given well structured, defined meaning, better enabling computers and people to work in cooperation.

Semantic Web is an extension of the World Wide Web with new technologies and standards that enable interpretation and processing of data and useful information for extraction by a computer. Hence semantic web is a web of databases and not of documents, queried by SPARQL [12].RDF m attaches metadata specify relations between the resources based on XML. Ontologies are another major semantic web technology built above RDF aims at providing strong semantics and vocabulary [13]. RDF link the data based on rational analysis, general dictionary and diagnostic thinking.

Semantic Web defines the relationships among [14] entities and Resource Description Framework, which is the language for labeling information and assets on the web. Inserting information into RDF files helps in making it possible for computer programs ("web spiders") to find, discover, analyze, choose, collect and process information from the web.

Figure4. Semantic web architecture

Resource Descriptive Framework (RDF) is the backbone of semantic web, which is the standard data model for data interchange on the web. RDF is for providing standard for metadata in semantic web. It is linked with the other resources via an identifier which uniquely identifies the subject. It breaks down the data into three parts namely resource, property and property-value [14].

Ontology describes basic concepts in a domain and defines relationships. Three main components of Ontologies are:

Classes or concepts.

Slots (sometimes called roles or properties) - attributes of each concept describing various features and attributes of the concept.

Facets (sometimes called role restrictions)-limitations on slots.

Ontology together with a set of individual instances of classes constitutes a knowledge base.

XML shorts for Extensible Markup Language which enhances the functionality of the Web by enabling user to identify user information precisely. In XML the semantics of tags and their associations are not fixed but can be defined as per application. In context of Semantic Web XML is the syntax of an conceptual schema or representation of data on the web, and is defined as a labeling tree. XML gives a formal platform to describe any grammar, so any data that have a grammar can be represented in XML. But XML only defines the structure of the document and doesn’t give any formal description of meaning or semantics of data contained in that document, so it lacks on achieving the semantic interoperability.

Semantic Learning Object Repository

Semantic Learning Object Repository (SLOR) is a web platform which stores learning object and their metadata, or only metadata, and provides rich interface to human or to software agents in searching and retrieving contents. The main drawback of LOR is that software agents (LMS) can’t understand the knowledge holding within the metadata.

SLOR provides formalized conceptual model using cognitive schema called Ontology. So by keeping Ontology of learning context, software agents can understand better the knowledge exists in the metadata domains. With Universal formalized conceptual model searching and exploring of learning content can be performed with reasoning and intelligence.

Figure5. Semantic Learning Object Repository Architecture

SLOR provides a set of suitable functionalities on each individual learning object conceptualization. The flexibility provided by the SLOR enables LOR to store normalized or un-normalized information about a learning object.

As shown in Figure5. , ‘LO-author’ (learning object author) creates a learning object; End user can use it by applying several repository searching mechanisms. An author can define a learning object in various forms, like creating a IEEE LOM metadata record including navigation control, the listing contents including table record, various record types and other information fields. Various LOM or Dublin core standard based metadata can be described which can refer to same LO. So in SLOR different Meta description can exist for same LO, this feature provides interoperability principle to external

Agent enables interoperation between different metadata descriptions to use same LO. Annotator does the work of automatic generation of metadata and meta-metadata [8].

Comparative analysis of Learning Object Repository and Semantic Learning Object Repository –

Table 2: Comparative analysis of LOR vs. SLOR

Key Functionality

Learning Object Repository

Semantic Learning Object Repository

Key Users

Learners ,Training Dept Admin, Instructors, Content Developers

Subject Matter Experts (SME),External agents, Annotator, LO creator

Key Services

Course Publishing & Enrolment, Author & SME Course Creation, Indexing of LO with Metadata

Creation of Ontology schema, Normalization of LO contents, Indexing of LO with Meta-metadata[8]

Level of conceptualization

LOR lacks the universal conceptual model to define what a LO is and what its metadata are.

There is a conceptual model defined that gives a formal definition about LO and its metadata

Interpretability

Knowledge transaction between different kinds of repositories cannot be implemented

SLOR provides restriction and interchange schema, so knowledge transaction can be implemented between different kinds of repository.

Runtime Semantics for LO

Standards are descriptive here so it gives only information or format about the learning content.

Standards are normative here, so it gives run time semantics of learning management system that uses LO.

Metadata information

Preventing agents to behave accordingly in response to the unstructured metadata written in natural language.

Agents behave properly in response to the metadata record written using knowledge base reasoning

Reusability

The composition of new learning materials as an aggregation of others is present here but at a limited level.

A knowledge base enables intelligence in aggregating different existing learning content, to make a new one.

Flexibility

Inability of adding new schemas to repository, to make metadata record classified according to the student knowledge levels.

New schemas can be added suitable for requirement, and there is a common vocabulary for LO definitions ,so it can fit all existing conceptualizations

Automation

Only manual browsing and searching are provided here without any reasoning and intelligence.

Functions can be delegated to automated systems by the existence of ontology schemas.

Conclusion and Future work direction

This paper discussed various pros and cons of LO and LOR. LOR provides query and interactive interface to user and software agents, facilitating rich search. Various E-learning architectures were discussed and analyzed. Non standardization of LO conceptual schema hamper knowledge interchange between repositories. Semantic Learning Object Repository is the best solution which enriches LOR by associating semantics. Semantic LOR are also facing many interoperability issues and user Relevance [15]. Information retrieval process using relevance feedback can be one solution for SLOR.