Most honorable chairman, esteemed committee members, DEAR family, friends and colleagues I welcome you to the public defense of my PhD thesis. In the following presentation I introduce you to my dissertation.
First, I give you short a introduction with two different examples to organise the world about us: we’ll look at a human-made way and so called self-organisation. I then introduce the context of this dissertation and the problem statement, following which we look at the hy*pa*the-sis and the structure of the study. I conclude with some lessons learned.
Humans create structured, hierarchical ways to organise the world around us, including different object and data. Already in the 3rd century BC the books in the Library of Alexandria were c atalogued. This system proved indispensable to librarians for centuries, and has influenced the art ever since. The idea is that information can be gathered and sorted-out overseen by a rational thought, and later accessed using that same rational path.
The concept and creation of the first cataloguing cards occurred during the French Revolution (1789)- The idea is that the object, for example a book, can be described and indexed separately by using metadata. Essentially, metadata is “data *about* the data”.
A cataloguing card represents metadata about each object in the collection and helps accessing the object. The metadata description determines *click* the author, *click* the title and *click* makes the access points available. The metadata description is usually done by librarians or trained indexers who know the cataloguing system, are familiar with controlled vocabularies and know the content area.
In nature, however, we can see different ways where order and patterns emerge from the bottom-up. Like these birds, no one is in control to tell them what to do, every individual follows 2 simple rules: “stay aligned” and keep the distance. When focusing on the colony level, *and not on individuals,* complexity and patterns arise.
An ant colony is an often used example where larger scale order emerges from small scale interaction. Self-organisation here represents the idea that even if individuals follow simple rules, the resulting group behaviour can be surprisingly complex and effective, for example, success in finding the shortest route to a food source.
Similar behaviour can also be observed among humans, we also look for the “shortest way for the food sources” and follow trails from others. In that process we leave marks, *Click* in this case foot paths emerge *Click* that are more efficient and time saving than the original road. it is typical that they are reinforced over time. *Click* This phenomenon is known as stigmergy.
Now I present the learning resource portal where most studies for this dissertation took place.It is an access point for primary and secondary teachers in Europe to find learning resources. *click*. The resources come from more than 20 different content providers and digital libraries from around the world. The content is aligned to different national curricula and, of course, is also in different languages. *click*. What makes it possible for teachers to search learning resources on this portal is the metadata that is used to describe each learning resource, just like in our earlier example of libraries. A LOM based application profile with mandatory indexing terms from a multilingual thesaurus is used to facilitate the cross-language retrieval of resources. sqdsdS
There are a number of problems when searching learning resources across language boundaries. (click) I give you examples of our main problems. The learning resource description by a librarian or indexer in one context, let’s say a learning resource described with metadata in Finland based on the Finnish curriculum, might not yield relevant results for a teacher who is searching for resources in another country context, say, in the Netherlands. Second, the expert vocabularies by pedagogical and linguistic experts, even when made in collaboration with end-users, seldom translate well to the information seeking tasks of the end-users - teachers.This problem is propagated when taken to a different lingual and country contexts. Despite the multi-million euro efforts of making standardised metadata available for sharing and reusing learning resources, we find that there is still room for improvement.
Furthermore, related to the problematic, I give you an example of Google. It is the most common way for anyone to find information nowadays. It’s success lies in the PageRank algorithm, which, unlike in our case, does NOT build on cataloguing or metadata. It actually measures the relative importance of a website, within a set of websites. This relevance is based on hyperlinks between these websites. It is important to note that these hyperlinks are placed usually when the content is created and rely on the creators’ judgement on the importance of these links. * Click *Unfortunately, such link-structure rarely exist between learning resources in heterogeneous repositories, and even less often across different languages. Thus, it is im possible to assign similar numerical weights for learning resources in a multilingual context. This forces us to evaluate other options.
So, an idea came about to combine the two approaches that were introduced in the beginning. The learning resource portal with metadata represent the human-made top-down order. So, then, of course, we wondered if the behaviour of an ant colony could be of any help? Tagging came about around 2003 and introduced a new way for any individual to organise digital objects around them by using simple keywords, called tags. *click* On our portal, this means that we let teachers add simple tags and ratings to a resource when they find it relevant. sqdsdS
As teachers add tags to individual resources that they find relevant, new patterns start emerging at the “colony level” which modifies the environment of the portal. * Click * The tags and ratings now appear related to the resources. * Click * On the other hand, lists of the most bookmarked items and global tagcloud emerge creating patterns on the system level. sqdsdS
This, in turn, has a potential to modify the behaviour of other individuals, as they might be inclined to use the tag as a navigational aid for their own resource discovery process.
So the hy*paa*thesis for tise dissertation was formulated. *click* Following the idea of ants finding the shortest path to a food source, it was hy*po*thesised that tags and other annotations would make users more efficient on the portal. *click*. We also borrowed the idea of social insect behaviour, by hy*paa*thesising that the tags make the system more robust and flexible. A robust system for us means that even when one or more elements fail, the users can still perform their tasks. And flexibility in a multilingual context allows users and metadata to adapt to a changing environment.
*click*The work started with 2 exploratory studies to set the scene: *click* first, we studied how users tag and behave in a multilingual environment. *click* Second, we studied how learning resources were actually used across contexts, meaning, a user who comes from one country uses a resource that originates from another country, and that is in a language other than his or her mother tongue. *click* The hy*pa*thesis was studied in a trilogy of studies based on empirical data collected from users of the portal. *click* First we looked at SOCIAL SEARCH, *click* then metadata ecology where the interaction of conventional m e tadata and social tags was studied.*click* the third study on the tag-resource connection and how it creates more flexibility across contexts.
We found that tags create new link-structures between learning resources in different languages. These resources, *click* represented here by Slovakian, Finnish and Spanish resources and which reside in different servers in different countries, never had any links between them *click* before the users created the tag “interactive” that now connects them using the orange edges. sqdsdS
Similarly, thanks to tags, new connections between users from different countries emerge. *Click* here are users from Austria, Hungary and Finland, *click* who all have showed their interest through tagging on the same learning resource. sqdsdS
Lastly, I would like to thanks everyone who was involved in this research. *click* I thank European Schoolnet for giving me an occasion to conduct my PhD research alongside my work, an arrangement which I truly appreciate *click* I am grateful for Helsingin Sanomain säätiö, a Finnish foundation who gives stipends for Finnish researchers to study abroad, for the generous stipend for 2006-2007. *click* I also express my thanks for OUNL and Celstec for hosting my PhD work here as a “buitenpromovendus”, and I thank the Dutch School for Information and Knowledge Systems for accepting my study as part of their programme. I herewith thank you all for your attention!
my 10 minutes of fame
Public defence of Riina Vuorikari November 13 2009, Heerlen, NL
<ul><li>Background: different ways to organise the world around us </li></ul><ul><ul><li>Human-made </li></ul></ul><ul><ul><li>Self-organisation </li></ul></ul><ul><li>Context of the dissertation & the problem statement </li></ul><ul><li>Hypothesis & structure of the study </li></ul><ul><li>Lessons learned </li></ul>Presentation outline
<ul><li>Discovering learning resources </li></ul><ul><li>across language boundaries is challenging! </li></ul><ul><li>Main problem: the resource & its metadata description are made in a different context from where the resource is discovered and/or actually used </li></ul>Problem statement
<ul><ul><li>Learning resources </li></ul></ul><ul><ul><li>from </li></ul></ul><ul><ul><li>different countries </li></ul></ul><ul><ul><li>and </li></ul></ul><ul><ul><li>in different languages do not cross-reference via hyperlinks! </li></ul></ul>Not applicable! Problem statement
Add a rating New kid on the block - tagging Add a tag
Social tags make digital trails visible! by Stiphy
<ul><li>The self-organisation aspect of a social tagging system on a learning resource portal helps users discover learning resources more efficiently. </li></ul><ul><li>User-generated tags make the system, which operates in a multilingual context, more robust and flexible. </li></ul>Hypothesis
Setting the scene Trilogy of empirical studies What happens when users tag in multiple languages? Are learning resources used across contexts? 2. Metadata ecology more robust and flexibile system 1. Ecology of social search more efficient users 3. Tags connect con-tent across contexts more flexibility Structure of studies
Spanish resource Tag «interactive» Tags create link-structures beween content in different languages Finnish resource Slovakian resource
Austria Hungary Finland Resource that «Travels well» Tags create link-structures beween users in different countries
<ul><li>This opens novel ways for better cross-language discovery of learning resources, paving the way for recommender systems, social navigation and ranking in this domain </li></ul>Outcome
I would like to thank everyone involved in this research, especially...