Ontology-based learning content repurposing: the ALOCoM framework.

This article reports on the development of a framework for repurposing learning object components, more specifically components of slide presentations. Unlike the usual practice where learning object components are assembled manually, the framework enables on-the-fly access and repurposing of learning object components. In earlier work, we have developed an ontology that formalizes structural aspects of learning objects. In this article, we present a framework that disaggregates slide presentations into this ontology format and reassembles their components (e.g., definitions, references) into new slide presentations.

Introduction

Issues concerning learning object (LO) re-use and repurposing are currently among the most important research topics in the learning technology community (Duval & Hodgins, 2003). In many cases, we need specific parts of a LO instead of the LO in its entirety. In that case, a definition, example or illustration is repurposed by copy and paste in new and different LOs. This approach is non-scalable in terms of maintenance, since each time a component is copied, a new place is created that needs to be maintained. It is possible to re-use LOs in a much more sophisticated way if we can access the components of a LO and repurpose them on-the-fly. This requires a more innovative and flexible underlying model for LO components (Duval & Hodgins, 2003).

In earlier work, we have developed an ontology that is an abstract learning object content model (ALOCoM), defining a framework for LOs and their components (Verbert & Duval, 2004). The ontology provides an explicit definition of the LO content structure, formally specifying both LO component types and relationships between those components. In this article, we present a framework that uses the developed ontology for composing and decomposing slide presentations. The framework transforms existing LOs from their tool specific formats (MSOffice, OpenOffice.org) into a representation compliant with the ALOCoM ontology. In this transformation process, the framework disaggregates LOs and provides direct access to their components, enabling their reuse in dynamic compositions of new LOs.

In the next section, we briefly outline the ALOCoM ontology. In the third section, we present the transformation framework, while we elaborate on LO repositories in the fourth section of the article. In the fifth section, the aggregation process and the generation of new MS PowerPoint and OpenOffice.org slide presentations is described, and the sixth section illustrates a scenario applying the framework. Related work, conclusions and remarks on the future work conclude this article.

The ALOCoM Ontology

In earlier work, we developed the ALOCoM ontology as a generic abstract learning object content model for learning objects and their components (Verbert & Duval, 2004). Figure 1 illustrates this model. The ontology distinguishes between content fragments (CFs), content objects (COs) and learning objects (LOs). CFs are learning content elements in their most basic form, like text, audio and video. These elements can be regarded as raw digital resources and are uncombined with other elements. COs aggregate CFs and add navigation. Navigation elements enable structuring of CFs in a CO. Besides CFs, COs also include other COs. Finally, LOs aggregate COs and add a learning objective.

We defined content types for each of these components. We introduced CF types such as image, text, audio and video. For defining CO types, we investigated existing information architectures, like the Information Block Architecture developed by (Horn, 1998) and the IBM Darwin Information Typing Architecture (Priestley, 2001). These architectures define information types (e.g., concept, principle, task) and their building blocks (e.g., example, definition, analogy). As a starting point, we defined CO types and their structure using DITA concepts, since DITA is a recent architecture with rich documentation and online support (Priestley, 2001). Besides CF and CO types, the ontology defines LO types. For now, only a slide presentation LO type is defined. Finally, the ontology defines the relationships between the LO components. Aggregation and navigation relations are specified. Aggregation relationships between components are represented in the form of a has Part and its inverse isPartOf ontology properties. Navigational relationships are specified as a list that defines the order of components in a CO or LO. For more information about the ontology, see (Jovanovic, Gasevic Verbert, & Duval, 2005).

[FIGURE 1 OMITTED]

The ALOCoM Framework

Our main focus is on the development of tools for extracting/transforming LO content into ontology-aware content (we call these tools disaggregators) as well as for repurposing ontology-aware content in real-world applications (aggregators). The ALOCoM framework supports both the process of aggregating and the process of disaggregating LOs. The framework maps different tool specific formats into representations compliant with the ALOCoM ontology and vice versa, ALOCoM instances into tool specific formats. For now, the framework supports slide presentations. Figure 2 illustrates the ALOCoM framework.

Since the most popular tools for slide presentation authoring are MS PowerPoint and OpenOffice.org (Najjar, Ternier, & Duval, 2002), the framework focuses for now on slide presentations authored using these tools. The process of content disaggregation performed inside the framework has an OpenOffice.org (OO) or a MS PowerPoint slide presentation as its input. The slide presentation is parsed and disaggregated into clear segments (slides, paragraphs, lists, list items, images, diagrams, tables, etc.). Text patterns are applied to categorize these segments into more meaningful components like definitions, examples, references, introductions, and summaries.

Finally, components are described by metadata using the Automatic Metadata Generation framework (Cardinaels, Meire, & Duval, 2005). For more information about this disaggregation process, we refer to (Verbert et al, 2005). The annotated components are stored in LO repositories. Currently, there are two different LO repositories (LORs). These LORs are described in the next section. Once we have LO components available, these components can be retrieved and reassembled into new LOs. Export functions to MS PowerPoint, OpenOffice.org, SCORM (1), HTML, PDF and PS are provided as illustrated in Figure 2. The fifth section of this article elaborates on aggregating new LOs.

ALOCoM Learning Object Repositories

Two LO repositories (LORs) for storing ALOCoM components are presently available, serving different purposes and different communities. The first one is an ontology-based LOR that stores ALOCoM RDF/OWL instances. The ALOCoM framework uses Jena, a Java Semantic Web Framework (http://jena.sourceforge.net/) to manage this repository. Jena offers persistent storage models, which are continually and transparently persisted to a backing store. Persistent models can be maintained in a file system, or in a relational database. We query the LOR using RDQL, a query language for RDF. While not yet a formal standard, RDQL is widely implemented by different RDF frameworks. RDQL allows complex queries to be expressed concisely, with a query engine performing the hard work of accessing the data model (McCarthy, 2004).

[FIGURE 2 OMITTED]

ALOCoM components are also stored in an ARIADNE Knowledge Pool System (http://www.ariadne-eu.org/). The ARIADNE Search-and-Indexing Tool (SILO) can be used by teachers and authors to search for fine grained components like definitions, examples, references, images or single slides. Furthermore, any other application that wants to have access to ALOCoM components can search for components in the repository using the Simple Query Interface (SQI). For more information about SQI, we refer to (Ternier & Duval, 2005).

Aggregating New Learning Objects

LOs disaggregated in an ALOCoM format provide us with a flexible solution for repurposing LO components. LO components at different levels of granularity are available (CF, CO, LO). For instance, we can retrieve complete slide presentations at the LO level, definitions and examples at the CO level or just text fragments or images at the CF level. These components need to be reassembled in new LOs. Currently, all selected components are assembled in a new slide presentation. The framework supports MS PowerPoint, OpenOffice.org, HTML, PDF, PS, and SCORM output formats. For more information about the export functions to SCORM, PDF, PS and HTML, we refer to (Verbert, Jovanovic, Gasevic, Duval, & Meire, 2005). We will illustrate the generation of a new MS PowerPoint and OpenOffice.org slide presentation in the rest of this section.

Export of an ALOCoM slide presentation to a MS PowerPoint or OpenOffice.org slide presentation proceeds similarly. To generate OpenOffice.org slide presentations, we use the OpenOffice.org Application Programming Interface (API). This API is a comprehensive specification that describes the programmable features of OpenOffice.org (http://api.openoffice.org). MS PowerPoint presentations are generated using the MS PowerPoint .Net API. The generation of PowerPoint and OpenOffice.org Presentation Objects is analogous, as manipulating and interacting with these objects is similar. In both cases slides can be added to a presentation object and a slide object is a collection of shapes. This is a presentation-oriented representation of a slide, in the sense that every component that is placed or inserted on a slide is a kind of shape with presentation properties. Structure related information in ALOCoM components is mapped to the aforementioned presentation elements of the OpenOffice.org/MS PowerPoint format. For instance, the title and the body of an ALOCoM slide are mapped to two different rectangles in a slide object. Since we do not keep track of presentation related information, we use default presentation styles for the title of a slide, list items and other components.

Scenario

We use the framework to repurpose existing LO components in new LOs. Currently, we enable uploading of both MS PowerPoint and OpenOffice.org slide presentations. All components of these presentations are available for repurposing.

A typical usage scenario of the proposed framework goes as follows. Suppose an author is creating a slide presentation on differential equations. He/she wants to start with a definition, followed by three examples. The author enters "differential equations" as keywords and selects "definition" and "example" as types of components that he/she is interested in. The system then searches the LOR and retrieves all components of the selected types dealing with the selected topic. The author chooses the most relevant components from the set prepared for him/her. Furthermore, the author wants to include a reference to a book (s)he wants to recommend and an image of the book. Again the author searches the LOR and selects the component he/she wants to repurpose from the set of retrieved components. The author is free to choose the presentation form of the generated content assembly among MS PowerPoint, OpenOffice.org, HTML, SCORM, PS and PDF formats. The author enhances the automatically generated slide presentation with some additional information on the topic, and the presentation is finally ready for in-class use.

Related Work

The TRIAL-SOLUTION project is developing tools to create and deliver personalized teaching materials that are composed from a library of existing documents on mathematics at undergraduate level (Lenski & Wette-Roch, 2001). Analogously to the ALOCoM work, the TRIAL-SOLUTION project defines an ontology for LOs that includes mathematical categories like a definition, theorem, proof, or example. The focus of the project is on document (de)composition and exchange of LOs for reuse. The TRIAL-SOLUTION System contains a splitter that decomposes document source files into a hierarchy of slices. For this decomposition, the presentation style of a particular author is taken into account. Also, it takes care of counters and key phrases assigned by the author. As such, the methodology for decomposing LOs is more accurate but less scalable than the ALOCoM methodology presented in (Verbert et al., 2005).

The dynamic Learning Content Management System (dLCMS) project is implementing a component model that is similar to the ALOCoM model. The dLCMS component model distinguishes between assets, content element, and learning units (Schluep, Bettoni, & Guttormsen Schar, 2005). Assets are media elements such as images, videos, animations and simulations and are equivalent to ALOCoM content fragments. Content elements can be associated to ALOCoM content objects as they are aggregations of assets. This component type is based on didactic content types, such as examples, exercises, self-assessments, etc. The dLCMS component model imposes more specific constraints to this component type. Finally, learning units aggregate content elements and can be associated with ALOCoM LOs. With the implementation of this component model, functionalities for flexible aggregations of content elements to learning units are provided. The project is not concerned with the disaggregation of LOs however, and assumes the availability of fine grained components.

CONCLUSIONS

In this article, we presented the ALOCoM framework as a solution for repurposing LO components. The framework enables repurposing of LO components (e.g., definition, example, reference) in existing slide presentations. Furthermore, these components are automatically reassembled into new LOs and launched in the authoring tool the author is using. Next steps will extend the framework to support MS Word and OpenOffice.org Text Documents. This extension needs to incorporate some additional component types that are not used in slide presentations, for instance different levels of headings. Furthermore, the efficiency and effectiveness of this approach for LO repurposing will be evaluated. This work will then result in a general framework for reusable LOs, that allows not only automatic repurposing of LOs, but also their components and that will enable the dynamic generation of LOs, adapted to the needs of learners.

References

Cardinaels, K., Meire, M., & Duval, E. (2005). Automating metadata generation: The simple indexing interface. In Proceedings of the 14th International World Wide Web Conference, Chiba, Japan, 2005, pp. 1140-1141.

Duval, E. & Hodgins, W. (2003). A LOM research agenda. In Proceedings of the twelfth international conference on World Wide Web, Budapest, Hungary, 2003, pp. 1-9.

Horn, R. E. (1998). Structured writing as a paradigm. In Instructional development: The state of the art. Englewood Cliffs, N.J.

Jovanovic, J., Gasevic D., Verbert, K., & Duval, E. (2005). Ontology of learning object content structure. In Proceedings of the 12th International Conference on Artificial Intelligence in Education, Amsterdam, Netherlands, pp. 322-329.

Lenski, W. & Wette-Roch, E. (2001). The TRIAL-SOLUTION approach to document re-use principles and realization. In Proceedings of Electronic Media in Mathematics, 2001.

McCarthy P. (2004). Use RDF models in your Java applications with the Jena Semantic Web Framework. Retreived from http://www.-128.ibm.com/developerworks/java/library/j-jena/

Najjar, J., Ternier, S., & Duval, E. (2002). The actual use of metadata in ARIADNE: An empirical analysis. In Proceedings of the 3rd Annual ARIADNE Conference, Leuven, Belgium, pp. 1-6.

Priestley, M. (2001). DITA XML: A reuse by reference architecture for technical documentation. In Proceedings of the 19th Annual International Conference on Computer Documentation, Sante Fe, USA, pp. 152-156.

Schluep, S., Bettoni, M., & Guttormsen Schar, S. (2005). Modularization and structured markup for learning content in an academic environment. In Proceedings of the PROLEARN-iClass thematic workshop on Learning Objects in Context, Leuven, Belgium.

Ternier, T. & Duval, E. (2005). Interoperability of repositories: The Simple Query Interface in ARIADNE. In Proceedings of the PROLEARN-iClass thematic workshop on Learning Objects in Context, Leuven, Belgium.

Verbert, K. & Duval, E. (2004). Towards a global architecture for learning objects: A comparative analysis of learning object content models. In Proceedings of the 16th ED-MEDIA 2004 World Conference on Educational Multimedia, Hypermedia and Telecommunications, Lugano, Switzerland, pp. 202-209.

Verbert, K., Jovanovic, J., Gasevic D., Duval, E., & Meire, M. (2005). Towards a global component architecture for learning objects: A slide presentation framework. In Proceedings of the 17th ED-MEDIA 2005 World Conference on Educational Multimedia, Hypermedia & Telecommunications, Montreal, Canada, pp. 1429-1436.

Verbert, K., Jovanovic, J., Gasevic D., & Duval, E., (2005). Repurposing Learning Object Components. OTM 2005 Workshop on Ontologies, Semantics and E-Learning, Agia Napa, Cyprus.

Notes

(1) Sharable Content Object Reference Model

Acknowledgements

We gratefully acknowledge the financial support of the Katholieke Universiteit Leuven research council through the BALO project and of the European Commission through the ProLearn Network of excellence, which has facilitated the collaboration between the Computer Science Department of the K.U. Leuven, the GOOD OLD AI group at the University of Belgrade and Simon Fraser University Surrey (SFU). The research of SFU is also funded by LORNET, a NSERC Research Network.

KATRIEN VERBERT, ERIK DUVAL, MICHAEL MEIRE

Katholieke Universiteit Leuven, Belgium

[e-mail]

[e-mail]

[e-mail]

JELENA JOVANOVIC

University of Belgrade, Serbia

[e-mail]

DRAGAN GASEVIC

Simon Fraser University Surrey, Canada

[e-mail]