1,841
edits
Changes
no edit summary
This part of the system, however sophisticated it may be, corresponds to what in the classic sense is "information retrieval" (IR). But it is a well-known fact that classic IR starts from a set of not very realistic assumptions. One of them is the idea that the user of the system knows exactly what he wants to "re"-trieve.
But what does he do when he has little idea, or no idea at all, of what he can flnd find in the system, especially so if the corpus is so vast and universal that it would take pages of text to indicate him the content (which, for this reason, he would not be prepared to read in advance).=== generator of animation and serendipity ===
This is why a second way of approach must be offered by the system as well.
Thanks to the openness of such a descriptive terminology, and a special cluster technique allowing for <big>fuzzy concepts and associative approach</big>, the system, would (secondly) also function as a generator of animation and serendipity in the processes of searching and finding. Such a facility will be a special, long desired feature and is well beyond the reach of a normal thesaurus. It may be worth remembering, that animation, as well as information that has not been called for explicitly, have turned out as the two long "neglected children" in accessing information.
The EMN project, however, has indicated, that even all this is not enough to fully cope with the problem of access to very large collections. On top of all this, there must be a third way of access to the known or unknown parts of the collection, and this may be driven by emotions and spontaneous action.
This facility is a real "hot" thing in two respects - one, as the system, in a way, would be capable of adopting the way the unknown user thinks, so that it acts as a learning system; the other as it allows the information provider to track the users individual approaches to the bank (of which nothing would otherwise ever become known), and to learn about their way of thinking, their very process of approaching the bank and of acquiring knowledge, and thus to be able to further improve the system.
=== constantly optimises the user-machine-corpus relation ===
So, the system will behave in an adaptive and "dynamic" way, adaptive insofar as it constantly optimises the user-machine-corpus relation, and dynamic insofar as, after a certain period of time, it will be more and more in itself a subject, agent, - and testimony -, of cultural change and interchange. Thus, it will more and more reflect the general thinking and paradigms of the different user communities and cultures all over the world.
Of course, the success of such a system heavily depends on a good input, i.e. the quality not only of the pictures collected, but also of the descriptive texts (captions, and other picture-related word material) coming along with the pictures.
=== A homogeneous text base === A homogeneous text base is a prime condition for the construction of the thesaurus which has to be built on the basis of the term material presented in the captions and other picture-related text material.
This is why, from the very beginning of the collecting activities, utmost stress has to be put not only on a top quality of the pictures, but also of their respective texts. These descriptions must strictly follow a matrix of informative and linguistic requirements, and this matrix, including a number of sample texts, has to be set up beforehand. The validity, as well as formal quality of both, pictures and texts, must be continuously monitored in the course of the operation.
The EMN project also has shown us the necessity of a homogeneous data input which right from the start must be geared to the requirements dictated by the access method(s) envisaged. If contribution is from various Museums, solutions in distributed, and networked data input, its update and control, are of prime importance. Also, there must be a build-up facility for individual picture collections with the different contributors.
It is totally unknown, yet, and impossible to predict, how the future visitor of the '”Virtual Multimedia Museum" will behave, and especially so what sort of use he will make once all these instruments of approach mentioned above will be within equal reach to him, and can be combined with each other.
To trace this behavior, therefore, will not only be an interesting thing in as much the system will reveal us something of how the user thinks and feels, acts and reacts, in the searches; It will also be most meaningful insofar as it enables to build a self-learning system on the grounds of this user-machine-corpus interaction.
