8. Implication for Toolbox Development

In this report we have shown how we have modelled case9c with IMS LD tools and LAMS. We have also discussed the problems identified by other research groups during their use of the specification and while developing their own solutions. Furthermore, we have demonstrated how it is conceptually possible to identify reusable structures in a learning design that can be used in templates. We have also identified a few issues regarding the development process with templates based on IMS LD.

Previously, we demonstrated that IMS LD supported the first part of case9c where we only had video and a text area. When it came to the grouping part and the e-mail functionality of part two, we experienced some problems. Currently, the only e-mail feature in IMS LD is notification, used for asynchronous learning scenarios. If we want users to send emails to each other, it has to be a part of the VLE and this VLE must support IMS LD. Our grouping problem can be related to the RELOAD Learning Environment’s implementation of role-parts because of the way it allows us to implement multiple roles for a student within an act. Although the specification defines this functionality, it was impossible for us to implement it. Similar issues related to grouping were also identified by Caeiro et al. (2003) and Miao et al. (2005). The problem is that, at runtime, the UoL should open for rearranging groups. Currently this is not supported by LAMS either, but in version 1.1 they are addressing this problem: “If the sequence contains group activities that are meant to be set at ‘runtime’, the teacher is prompted to set the groups based on the current students in the class.” (LAMS Wiki). To be able to support true social constructivism and collaboration, this is a feature that the CALIBRATE toolbox LOs should provide. This means that a template should give the teacher the opportunity to create or modify groups at runtime. A possible solution could be to implement this functionality in JavaScript?, in case we want the LO to have a pure html-interface, or as a Flash object. LAMS, as an example, uses Java on the server side and Flash for the client’s interface. Similarly, our architecture is based on Python for the server side, and Flash or JavaScript? could be used for generating the interface.

The other problem we identified is that it is difficult in RELOAD, and IMS LD in general, to create complex activity structures. In case9c it was not too difficult as the structures were rather simple (and we had experience with IMS LD). This is something that Caeiro et al. (2003) also reported. In LAMS, on the other hand, there is an intuitive way of creating sequencing, but the transition between activities currently lacks advanced features. The LAMS development team is, however, working on more complex structures, like the parallel activity tool, and improved optional sequences and transitions (LAMS Wiki). Neither in IMS LD nor in LAMS were we able to model indefinite loops as they exist in the knowledge-building module of Fle3 and discussion group tools. That is probably related to the xml-binding of an IMS LD and the lack of properties in the current specification. If IMS LD had a vector datatype property where values could be stored as well as a loop mechanism, then the issue would be addressed. For this to work in an authoring mode, we would probably have to implement these transitions in a graphical way, much as we see in the Flash interface of LAMS. However, we could also let the transition go in a loop instead of just from one activity to another. The loop transition could be an arrow that points back to the activity that it departs from in order to indicate that this activity is a loop activity, or, an arrow that could point to an activity in an earlier stage of the sequence (Figure 15).

Figure 15: Showing a remedial learning loop where the student can do the same activity again.

A third problem we need to address is that IMS LD does not support artefact creation at runtime as reported by Miao et al. (2005). This is something that we could implement as tools or services which can be loaded by the LO at runtime, exactly when users need them. For example, we could provide artefact creation like polls, chat and argumentations as services bundled with a toolbox installation. However, there are potential interoperability problems due to role-parts within these services, but this could be solved through the use of usernames (a chat “username” for example).

A fourth issue regarding the toolbox is how we should handle roles. Students and teachers can take different roles in a learning scenario. In IMS LD, practice has shown that roles are too rigid in acts. There are no properties or conditions on how different roles can interact with each other. In SCORM the role-part is totally non-existent. If the students using a SCORM based LO (also known as SCO) are intended to take different roles, this would have to be specified in a document describing the pedagogical method of the learning scenario. The learner role is taken for granted and teachers have no way to intervene. Teachers can intervene if the VLE has support for that, but it is not a part of the LO. If the toolbox takes roles seriously, we should either implement IMS LD or make sure that roles are a part of the LO. And we should also make sure that a teacher has the possibility to intervene in the LO’s learning design at runtime. In LAMS this is implemented as an interaction between the LAMS UoL and the LAMS monitor interface, but LAMS is a VLE as well and is the only VLE that can use LAMS UoLs?.

The idea behind this work package was to try to create templates for LOs based on social constructivism and pedagogical methods identified in the CELEBRATE project. As we identified from the COLLAGE project, they had only a few templates for teachers to use. As reported in the analysis chapter, the teachers experienced problems when trying to assign resources to the template and they had problems understanding the pedagogical method embedded in the template. The toolbox should not have templates for only text plus images. Rather, these templates should be made in such a way that they have a clear pedagogical profile. This profile should be presented to the user graphically as in LAMS and there should be markers for collaboration, loop activities and sequencing. LAMS has markers for collaboration (G standing for Group) and sequencing (transition arrow). We should provide the same, but loop transition markers as well. If a template has a service or tool embedded, this should also be marked in the interface (e.g. a Fle3 activity logo).

8.2 Practical problems using IMS LD

The CELEBRATE project identified that there are currently a wide range of VLEs, LCMS and CMS in educational settings in Europe (Vuorikari, 2003). In the final report (McCormick? et al., 2004) there is a discussion whether the EUN portal should be a VLE rather than a Learning Resource Exchange but, at present, it looks as if the LRE portal will initially primarily facilitate the exchange of LOs. Hence an important question is whether the toolbox should deliver LOs in IMS LD format.

It is likely that current VLEs in Europe will not be able to run these LOs in IMS LD format. These VLEs do not have the runtime engine to interpret the structure of activities in a UoL. A solution to this problem could be to opt for a JavaScript?- or Flash-based object that embeds activity structures, conditions and properties. The toolbox LOs could provide more conditions and properties than IMS LD currently have in order to increase the possibility to represent e.g. remedial learning (loops). If the toolbox creates an ordinary LO with its own JavaScript? or Flash object, all VLEs can interpret this LO, because it basically uses JavaScript? or Flash ActionScript? to refer to resource files that are compressed within the LO. We do not have to use the runtime model of SCORM because not all VLEs can interpret information from the runtime model and currently EUN schools do not need personal portfolio information or test scores.

IMS Content Packaging v1.1.2 is one good option. The IMS Content Package (IMS CP) consists of two major elements: a special XML file describing the way content is organized and referenced resources, and the actual physical files being referenced by the XML. The special XML file is called the IMS Manifest file, because course content and organization are described in the context of 'manifests' and everything is compressed together using PKZip v2.04g. Its organisations element is different than an IMS LD UoL where the learning design is a part of the organisations’ element replacing the simple hierarchical structure of an IMS CP v1.1.2 compatible LO (Figure 16). Using IMS LD CP v1.1.2, the VLE can start out with the toolbox LOs first file and the rest is handled internally by the toolbox LO (e.g. through JavaScript? or Flash). For the VLEs that do not support IMS CP v1.1.2, we must provide information so that they know which file is the default one, which usually is the first file referenced in the resources element.

A potential problem if we do not use IMS LD is that it will be difficult to capture roles and group dynamics, which are central in collaborative learning. Usually this is handled by the VLE by reading the different role-parts from the UoL (if IMS LD compatible). However, it should be possible to create LOs which embed a few features that normally would be in a VLE. If we use Flash based LOs we could embed features for role-parts by (as mentioned earlier when considering artefact creation) providing usernames when entering the LO (like a nickname or username). The teacher or administrator creating the session for using the LO could assign the teacher’s role, but leave the LO open for the students to organise themselves within the LO. Then these usernames could be used when accessing other services and tools as well. The premise of this solution is that EUN schools do not want to store results of student portfolios and tests. In this way we do not put any pressure on the VLEs to handle role-parts or group dynamics like in the IMS LD specification, but rather as a feature that the toolbox templates and toolbox LOs provide.

Figure 16: Top half shows the IMS CP v1.1.2 where the organizations element differs a lot from IMS LD Content Packaging on the lower half.

8.3 Conclusions and recommendations

In this study we have shown that there are both advantages and disadvantages of using IMS LD for creating pedagogical templates in this project. For the CALIBRATE project we will show the advantages and disadvantages in a structured list:

Advantages:

- IMS LD is based on thorough analysis of contemporary pedagogical research

- IMS LD's structures for separating roles, content and activities are important if we want to create pedagogical templates.

- There exist pedagogical templates for advanced learning models based on IMS LD.

- IMS LD is capable of supporting well and predefined structures of collaboration as in case9c.

- IMS LD is more capable of supporting advanced pedagogy than other influential specifications such as SCORM.

Disadvantages:

- IMS LD does not have structures or properties to support remedial learning loops or iterative processes as in “Progressive Inquiry Model” and the Fle3 environment.

- IMS LD does not have structures or properties to support dynamic group behaviour as in rearranging groups at run time.

- IMS LD does not have structures or properties to support artifact creation at run time.

- Most Virtual Learning Environments will not be able to interpret the structures and properties of IMS LD and thus will not be able to run “Units of learning”.

Our conclusion is that the disadvantages outweigh the advantages of using IMS LD structures for creating pedagogical templates in this project. There are two main reasons for this. Firstly, most VLEs in Europe today will not be able to make use of a learning resource built on IMS LD. Although IMS LD as a specification and framework will be interesting to use in the future, we must find other ways in which roles, content and activity descriptions can be exported in one package from a learning environment in Europe to another elsewhere. Secondly, our study show that although IMS LD is able to express collaborative learning scenarios that are well structured in advance (like case9c), it is not able to express loosely-structured processes such as are seen in the Progressive Inquiry Model. What we recommend to CALIBRATE toolbox developers is that they:

- Divide between content, pedagogical methods (activities) and tools. The reason is that activities will be reusable without being too closely connected to content and external tools. Then teachers can take an existing activity and add content and tools as they see fit.

- It must be easy for teachers to understand the pedagogical methods of a template. When a teacher wants to select a pedagogical template, it must be clear and intuitive how to use the method this template represents.

- To be able to run advanced activities in Virtual Learning Environments within Europe today, we must opt for client side script languages such as Flash or Javascript. The reason for this is that it would hinder interoperability if we would try to transfer Learning Objects from one system to another based on server side scripting languages or programming languages such as Python, PHP or Java.

- To be able to run advanced activities from a Learning Object in Virtual Learning Environments within Europe today, we must use the programming power of Flash and Javascript. These two client side script languages can provide reasonable dynamics regarding the flow of activities in a learning session (e.g. If we want to create activity loops), artifact creation at run time and group behaviour. Export Learning Objects as IMS Content Packages that reference compiled Flash files, or Javascript files. The reason is that all learning environments in Europe can read IMS Content Packages.

- We should avoid referencing external tools and environments in an IMS Content Package, as this hinders interoperability. If external tools and environments are referenced, teachers and students need to have access to these external tools and environments. This could cause problems. Instead, the pedagogical templates should embed most of the methods seen in these external tools and environments.