WP1 Improving the Semantic Interoperability of Learning Resource Descriptions
Getting the tags right...
The number of resources available on the Internet is staggering but teachers often can’t reach the ones they want. Although each resource is tagged with information (metadata), they are faced with the two-fold problem
that digital learning resources are described in local languages that they don’t necessarily understand and
that the current metadata descriptions are not precise enough to be of use.
In CALIBRATE we attempted to research and test approaches that could improve the discovery and evaluation of digital learning resources. What we did is to create vocabularies that represent the key terms found in different European national curricula in such a way that computers could read them. Since the terms used in these vocabularies represented the semantics of the national curricula and existed in all languages, they would add precision and overcome some of the language barriers when searching or browsing for learning resources.
In addition, we constructed a novel way to evaluate and discover learning resources: teachers can link learning resources to a part of their own curriculum which in turn links it automatically to the parts of curricula in other languages where the semantics is roughly the same.
Creating a new vocabulary
What we wanted to achieve was the creation of a vocabulary that could represent the descriptions of meaning and the intended use of a digital learning resource so that it could firstly be discovered and then evaluated. We also wanted to ensure that this vocabulary was harmonised for all languages in the EU. This work could be seen as one part of the infrastructure to improve metadata tagging and discovery services in the CALIBRATE Learning Resource Exchange Portal (LRE Portal).
Both providers and users of digital learning resources need access to such a set of subject keywords, harmonised across Europe. In order to start addressing this issue, the CALIBRATE project chose to use the different national curricula as a starting point. The idea was that we create the vocabulary based on terms that we identify from the different national curricula and that these terms cover the semantics found in these curricula. To be able to do this we made an extensive analysis of eight curricula from four nations for the subjects of Mathematics and Environmental Studies.
Each term we identify from these curricula must exist in all languages and be semantically represented in most curricula. By “semantically represented” we understand that the meaning of a sentence or section in a curriculum can be represented by one or many terms. These vocabularies are then created in such a way that we can automatically map a term in Polish for example to the equivalent term in German or another language by using English as a spine vocabulary. For example, we worked on vocabularies for topics and action-verbs (1) (educational objectives) that together represent the targeted competencies (hence the semantics) found in the curricula.
When a publisher describes the intended use and meaning of his/her digital learning resource, using IEEE Learning Object Metadata (2) (IEEE LOM), s/he will, in addition to the other metadata, use one or many topics/subtopics and one action-verb from the controlled vocabularies that we have created. The additional metadata that we have added (topics and action-verbs) are placed within the standardised way of representing metadata so that learning resources can be transferred from one learning resource repository to another without any complications. In technical terms this means that, if the vocabularies for topics and action-verbs actually represent the meaning found in national curricula and that each learning resource repository in the CALIBRATE project is able to read the additional metadata of topic and action-verb terms (which they should), we will have identified one way of solving the problem of semantic interoperability.
Developing a curriculum mapping tool
Another part of our work was the creation of a Curricula-mapping tool (3). This pursues the same aim - the improvement of the infrastructure for metadata tagging and discovery services in the LRE Portal. The purpose of this tool was to connect the terms in the controlled vocabularies (topics and action-verbs) to the different national curricula so that we could create associations between the targeted competencies (topics plus action-verbs) and sentences in the curriculum. The idea was that we wanted to use these associations when teachers evaluated and discovered learning resources in the LRE Portal. This was a novel approach: a teacher would be able to link a learning resource to one or many sentences of his/her own curriculum to which s/he thinks it is suited (Figure 1).
Figure 1: Linking a resource to a curriculum sentence. The resource isbe linked to the sentence “1.1.1 Concept formation/knowledge of facts” and the association used is “Associate rational numbers with real and meaningful context”.
Because of the associations connected to these sentences, the learning resource is automatically linked to sentences in other countries’ curriculum that have the same associations. So, when a teacher from one country browses in his/her curriculum for learning resources within a subject domain, s/he can find learning resources not only from his/her country, but also learning resources from the other countries as well. This means that a teacher from Poland can find Czech learning resources when browsing the Polish curriculum for a certain topic (Figure 2).
The last issue of linking learning resources to a curriculum and browsing a curriculum to find learning resources was the last to be validated. We did not know whether teachers found it practical to use their national curricula as a way to find learning resources. We also did not know to what degree they would find it practical to use learning resources in other languages. This was partially evaluated during the CALIBRATE Summer School in August 2007.
Figure 2: Browsing a Math curriculum. There are learning resources linked to each sentence, and the learning resources target the associations for each sentence. These learning resources could be from learning resource repositories all over Europe.
The action-verb vocabulary equals Bloom’s extended taxonomy. We also worked with other vocabularies derived directly from European curricula that can replace the action-verb vocabulary if necessary. (up)