This poster presents a practical demonstration of the current iteration of CoFIND (Collaborative Filter In n Dimensions), a web-based collaborative learning environment, the structure and content of which is generated by the combined individual actions of the learners who use it.
CoFIND is predicated by the assumption that there are many useful learning resources available on the World Wide Web, that learners will be able to find some such resources then learn from them through a process of study and conversation. The main module which defines CoFIND is a system for collaboratively bookmarking learning resources. Links to resources are entered by the learners themselves and are stored as URLs. These URLs need not refer to an online resource as the system allows learners to generate and store a web-accessible textual description of any resource at all, be it a film, a book, a person, a place or a particularly interesting object.
Incorporating a collaborating rating mechanism, CoFIND allows its users to generate new metadata about learning resources in the form of evaluation criteria known as qualities. Once a learner has entered a new quality, all resources may thenceforth be evaluated by all users of the system according to how well they match it. It is CoFIND's ability to rate resources in more dimensions than the conventional like/dislike or use/not-use criteria of a conventional collaborative filter that provides the n dimensions referred to in its name.
Qualities are the features that learners value in resources. Typical qualities are "useful" , "good for beginners", "comprehensive" and "easy to understand", although learners are free to enter any quality that they wish. Resources are returned in a list ordered by how closely they match a selected quality, using a metric based on rating and popularity (measured by the number of ratings a resource has received). It would be easy for the qualities generated by learners to grow into an unmanageable mass, much of which would be of little value to the bulk of the learning community. Therefore, the list of selectable qualities is ordered by popularity, with unpopular, seldom used qualities falling off the bottom of the list altogether in a process analogous to evolution. Successful qualities thrive by providing lists of resources that are useful to learners.
As well as its collaborative bookmarking system, CoFIND incorporates a number of more conventional features found in online learning environments, including the ability to upload files to the system and the means to discuss any resource using a fairly conventional asynchronous discussion mechanism. Uploaded files and individual discussion messages may be added to CoFIND as resources, allowing the collaborative rating system and discussion mechanism to act as ways of evaluating and assessing resources generated by the learners themselves.
CoFIND allows topics (distinct from qualities in that they provide binary categorisations of resources into subject groupings) to be entered by the learners, thus creating a collaborative course structure. Work is still in progress to find a way make these topics self-organising, but at present they are simply accepted by the system and allow their originators to place them into a hierarchical structure.
The combination of collaborative rating in n dimensions, collaborative resource generation and categorisation, together with the means to discuss and comment on resources gives each CoFIND system the ability to build and to organise itself according to the inputs of its users. No CoFIND system is quite like any other and each CoFIND system can organise itself in many different ways depending on the qualities that learners are seeking in resources, and so it tailors itself to the specific needs of each individual learner and each group of learners that use it.
CoFIND is implemented using Microsoft's ASP. It is configurable to allow its administrators to limit the ability of learners to add topics, qualities, resources or upload files. This gives it the flexibility of behaviour to enable it to be used in a range of more conventional environments than the completely self-generating form of the full version.
CoFIND has been implemented in a number of different forms to provide support for conventional taught courses, including the following examples:
This poster demonstrates these systems and the results achieved through using them.
Links to various CoFIND systems and related resources may be found at http://www.it.brighton.ac.uk/staff/jd29/cofind.html