<?xml version="1.0" standalone="yes"?>
<Paper uid="W01-1508">
  <Title>for Multimedia Language Resources</Title>
  <Section position="6" start_page="1" end_page="1" type="concl">
    <SectionTitle>
4 Tools and Metadata
</SectionTitle>
    <Paragraph position="0"> It was the initial idea that the metadata descriptions could also have elements that describe specific tools that can be used to act on the resources themselves. However since resources and tools form orthogonal dimensions it is better to have the metadata description only describe resources and not a set of tools that will change in time anyway. A more elegant solution is to describe the type and structure of the resources in sufficient detail so that &amp;quot;browser&amp;quot; tools used to access the metadata description can decide which ones of the available tools are suitable to handle the data. This can be either based on local user configurable information or on some sort of remote tool registry. At the moment IMDI is experimenting with a scheme of (semi) mime-types to characterise language resources. We foresee that users will want to customise the mapping of tools to resource types to their own taste just as they are able to do with WWW-browsers.</Paragraph>
    <Paragraph position="1"> Needed for such a scheme is that tool repositories note the types (mime-types) and encoding, character encoding for which the tools are suitable (see the lists in table 1 and 2). It has to be investigated in detail how far tool registries and resource collections structured by metadata descriptions can be created in a way such that especially naive users can overcome the frustrating problems of accessing the right resources with suitable tools. This problem is not solved and is one of the greatest obstacles for increasing the reusability of the huge treasure of resources. IMDI has taken limited tests with a number of tools to study the interaction between mime-type tagged resources and selecting from a tool palette. We have no doubt that this is the way to go.</Paragraph>
    <Paragraph position="2"> A special question is the form of the infrastructure. Where will we store the metadata descriptions and/or resources and how are tool registries such as from DFKI [13] made known to the distributed resource universe? During the IMDI project a preliminary solution is found for creating a registry authority for metadata. This registry authority has to build a web-portal, check the quality of the produced meta descriptions, create intuitively understandable browsable hierarchies based on the meta descriptions and link the meta descriptions to other type of information and resources The registry authority will also provide tools such as a constrained editor that allows the user to create meta descriptions and a suitable browser which can operate on the metadata description files.</Paragraph>
    <Paragraph position="3"> The IMDI project will also work on requirements for the registry authority and the metadata tools.</Paragraph>
    <Paragraph position="4"> At the moment the time has come for IMDI to investigate if and how the metadata description browser can access remote software registries to assist users in the choice of tools to use for resources. This would be a logical extension to the local configurable mapping of tools on resource types that is needed anyway for nonnetworked situations.</Paragraph>
  </Section>
class="xml-element"></Paper>