|
|
Retrieval of Education Materials on the World Wide Web:
Overview of Projects, Issues, and Directions--1998
WORKING DRAFT, v.0.5
Jacqueline Russom, Senior Researcher
Scholarly Technology Group, Brown University
(Jacqueline_Russom@brown.edu)
July 31, 1998
The
US Department of Education supports numerous public access Web sites that
provide information to educators, administrators, policy makers, and
students.
The WALDO project is one of several responses to an initiative calling for
the
provision of "one stop" access to materials on any of these sites. This
report,
part of Brown University's (Scholarly Technology Group) component of WALDO,
reviews the issues involved in information retrieval from the Web, with
particular reference to cataloging resources for the education community.
Recommendations specify ways to exploit features of existing metadata sets
and
authority lists to accommodate the full range of educational materials on
the
web.
The
growing body of educational material on the world wide web cannot be
located
efficiently at present by potential users [HRST], [DC1]. Two approaches to
this problem are currently being pursued: (1) searching content and (2)
searching specific bibliographical information or "metadata." In the first
approach, query terms are matched to terms in indexes created by a program
from
target documents. Documents whose terms match the query terms are
presented as
items of possible interest to the searcher. Such automatic indexing
programs
have been implemented by the Office of Educational Research and
Improvement's
Cross-Site Indexing project and by the Northeast and Islands Regional
Educational Lab [CSI], [LAB]. This approach typically presents the searcher
with far too many items, since a randomly selected word in a document is
unlikely to provide much information about its focus of interest. Many
relevant documents may not be recovered at all. Statistical methods can be
used to make better use of vocabulary information ([SAL], [IR]), but in the
absence of information that distinguishes terms of the title, author,
subject,
or other crucial roles from the uses of words in general, such methods are
limited.
It seems necessary, therefore, to pursue the second approach: to
characterize
the contents of web documents by means of an electronic cataloging
procedure.
The bibliographic information about a document required for efficient
retrieval
is known as "metadata." Metadata records contain information about a
resource
in a structured format as values of labeled attributes. This information is
much like that in existing catalogs such as the ERIC database [ERIC] or
ENC's
Resource Finder [ENC]. Metadata from a variety of sites can be harvested,
indexed, and then searched in response to a request for information that
refers to specific attributes or identifies values that are of interest.
All
and only the materials whose records are appropriate matches to the query
are
returned.
A
variety of metadata formats designed for similar purposes (MARC records,
the
GILS system, TEI headers, IAFA templates) are being used for bibliographic
control and other information management projects (see [H&D] and [RUS]
for
a review and evaluation of these systems). But many of these formats cannot
provide useful models for retrieval of educational materials because they
need
to be implemented by a staff of trained catalogers and are not suited for
the
huge number of documents on the web. Such formats are designed for
substantial
information management tasks not pertinent to retrieval of the educational
materials with which this report is concerned. One metadata model is
significantly more useful, largely because it is simple enough to be used
by
non-professional catalogers. This is the Dublin Core, a basic set of 15
metadata elements needed for retrieval of web documents.
The
original plan for the Dublin Core emerged from a workshop held in Dublin,
Ohio,
in March 1995. The goal of the meeting was to achieve consensus among
librarians, scholars, and internet standards makers on "a list of metadata
elements that would yield simple descriptions of data in a wide range of
subject areas" [DC1]. Subsequent workshops proposed design principles that
would keep the Core small and simple by specifying how extensions and
modifications could be handled by individual implementers. All Dublin Core
elements are optional and repeatable in principle, though they may be
otherwise
constrained by individual implementers.
A basic 'container' architecture was developed to provide for the
representation of additional metadata element sets. This "Warwick
Framework"
[DC-2] underlies the evolving Resource Description Framework that will
extend
the metadata model to accommodate specialized information management
requirements without altering the integrity of the basic Core.
The fourth metadata workshop, and subsequent discussions on the meta2 list,
have developed the notion of qualifiers as a mechanism to refine the
elements.
The "Canberra Qualifiers" [DC-4] extend the basic elements by allowing
them to
be qualified in three ways:
- Scheme: the identification of the authority list or standard from which
the
value of an element is obtained (such as an indication that subject terms
are
selected from the ERIC thesaurus, or that date formats follow the ISO 8601
specification)
- Lang: indication of the language of the content of the metadata element
(a
document in Spanish would have a DC.Language value of "Spanish", but the
Lang
qualifier would make it possible to indicate that a set of Subject values
was
in English).
- Subelement: a mechanism for more narrowly specifying the meaning of a
given
element (for instance, to indicate whether the value of the Date element
represents the date of creation of the document or the date that it was
cataloged, etc.). Qualification with subelements makes it possible to
characterize resources much more accurately, but it also introduces the
possibility that the metadata records will become too complex to be
created by
non-specialists and too divergent to be interoperable. The needs of
implementers varies and the model must work for both 'minimalists'
concerned
with ease of use and 'structuralists' concerned with accuracy of
description.
One of the major outcomes of the fifth and most recent workshop (Helskinki,
October 1997) was stabilization of the set of 15 Core elements. A plan was
drawn up for documenting both Simple and Qualified Dublin Core in a
series of
Internet Drafts in order to move towards formal standardization and
international acceptance. The first document [RFC1] has been submitted.
It
discusses the conventions for use of the basic Dublin Core elements, as
they
are described in the reference description [DC-REF], without qualifiers.
These
elements can be divided into three groups:
- Elements related mainly to the Content of the resource (Title, Subject,
Description, Source, Language, Relation, Coverage)
- Elements related mainly to the resource when viewed as Intellectual
Property
(Creator, Publisher, Contributor, Rights)
- Elements related mainly to the Instantiation of the resource (Date, Type,
Format, Identifier)
Four additional RFC documents will describe (a) conventions for embedding
unqualified Dublin Core metadata in an HTML file, (b) recommended
qualifiers
and principles for qualifying DC elements, (c) conventions for embedding
qualified metadata in an HTML file, and (d) a description for encoding
qualified DC metadata in compliance with the Resource Description Framework
(RDF).
The combination of the basic simplicity of the Dublin Core with a capacity
for
extension through use of qualifiers or collection-specific elements has
motivated implementation of DC metadata in a great variety of disciplines.
Implementers must determine for themselves how to balance the need for
richer
bibliographic description against the need for efficient cataloging.
ED-sponsored
Web sites of particular interest for the WALDO project are the Regional
Technology in Education Consortia (RTECs), the Regional Educational Labs
(for
research on school improvement), and the Eisenhower National Clearinghouse
for
Math and Science. The materials for educators at these sites are primarily
concerned with public schools (K-12), with some attention to programs for
vocational training and adult literacy. They include:
- curricular materials (lesson plans, descriptions of activities,
multimedia
presentations)
- documents (research reports, guidelines, regulations, policies,
professional
development materials, news)
- databases of information about off-line books, articles, films, etc.,
- lists of links (internal links to components of a document or project;
directories of external links to sites or resources on specific topics)
- software packages (tools for assessment, course development, etc.)
To
identify the kinds of metadata needed for cross-site discovery of
educational
resources, it is useful to consider the kinds of cataloging systems that
are
familiar to educators. The ENC Resource finder represents a familiar
system
for classifying curricular materials; the ERIC database is the
standard
for classification of documents and journals dealing with
education issues.
The Eisenhower National Clearinghouse catalogs and evaluates curricular
materials for mathematics and science education. Most resources cataloged
are
not in electronic form, and those that are available on the web cannot be
directly accessed, in that a search would take the user to the ENC Resource
Finder database rather than to the resource proper. Resource Finder
records
are created by expert catalogers and incorporate a rich classification
system
designed to facilitate ordering of materials and assessment of their
suitability for a given educational purpose. In addition to the kinds of
information represented in the Dublin Core set, these records contain
fields
for grade and audience, vendor, physical description, cost, product
identification number, national or local standards met, evaluators'
judgments,
and extensive description of the conceptual substance of the resource,
including abstract and table of contents. A decision needs to be made as
to
which of these many fields should be represented as metadata for cross-site
searching and which should remain as local information available once the
resource has been selected. The most obviously useful extensions to the
Dublin
Core for materials targeted by the Resource Finder would be elements for
grade and audience. An appropriate subject element
for
curricular resources would need to distinguish academic discipline
(English,
biology, etc.) from concepts or key terms that characterize the resource
more
narrowly (epic poetry, frogs, etc.).
The Education Resources Information Center has been abstracting and
disseminating educational publications for forty years. Publications are
cataloged by subject specialists at clearinghouses dedicated to specific
topics
of interest. The fields of the ERIC database provide a straightforward,
minimal, and intuitive bibliographical description. They could be
represented
as Dublin Core metadata without significant extension of the element set.
The
ERIC activities requiring specialists for implementation are creation of
abstracts or digests, assignment of subject descriptors from the ERIC
thesaurus, and determination of publication type. Creation of abstracts and
digests does not concern us here, as it is not directly relevant to
metadata
design. But even the simplest metadata system must confront the problems
of
selecting appropriate descriptive terminology. ERIC's thesaurus and list of
publication types are much too extensive for use by nonspecialist
catalogers,
incorporating detail that might in another system be represented as
discrete
fields of the record, such as Audience or Instructional Method. Metadata
developers must address the need for simpler subject and resource type
vocabularies dedicated to web search requirements. This might be
accomplished
in part by associating a somewhat wider set of elements with smaller, more
manageable vocabulary lists.
As
noted above, the Dublin Core has emerged as the dominant architecture for
classification of web materials. The major education metadata projects
analyzed below either were defined as Dublin Core projects from the outset
or
were adapted to the Dublin Core model in the course of development.
<
http://www.edna.edu.au/edna/owa/info.getpage/?sp=auto&pagecode=5210>
- Scope of project: EdNA is a collaboration between Australian states and
territories and all sectors of education and training: schools, vocational
education and training, adult community education, and higher education.
Cataloging of education resources is decentralized, being undertaken by
educational institutions themselves. Over 71,000 documents have been
cataloged.
- Materials being cataloged: Participating primary and secondary schools
have
so far cataloged resources that are mostly 'external'--i.e., descriptions
of
educational components of government or corporate websites, such as NASA
or the
Smithsonian. Vocational and higher education participants have cataloged
'internal' curricular and training materials.
- Metadata elements: The EdNA metadata record is based on the Dublin Core
but
does not use the Relation, Source, or Contributor elements.
- Most additional elements are related to the management of the records
themselves (meta-metadata):
- Entered (date of creation of metadata record)
- Approve (approver of item for inclusion)
- Suggestor (suggestor of item)
- Reassessment (months until resource should be reassessed)
- Categories (classification system in directory of resources)
- Two additional EdNA elements recognize the unique hyperlinked character
of web resources and specify constraints on the harvesting of documents to
index:
- IndexLevel (number of levels of links to follow)
- IndexSites (number of servers to access when following links)
- The only elements added to the Dublin Core set that apply specifically
to the educational character of the resources are:
- Review (third-party review of the resource)
- UserLevel (controlled vocabulary of user and school level)
Comment: The EdNA metadata set is among the simplest in use for
education resources. Subject terms are uncontrolled key words and the
vocabulary of users is limited to 'students' and 'teachers'. It is not
clear if this set will provide sufficient description for discovery of the full
range of resources for the education community.
<http://ariadne.unil.ch>
- Scope of project: Ariadne is part of the education and training program
of the European Union Telematics Application Program. Its goal is to
foster the share and reuse of electronic pedagogical material by
universities and corporations.
- Materials being cataloged: The Knowledge Pool System is a
database of reusable pedagogical materials and metadata records describing
these materials. It is currently distributed at eight sites throughout
Europe.
- Metadata elements: The Dublin Core elements and additional Ariadne
elements are grouped into several categories:
- General information about the resource (Dublin Core Identifier, Title,
Creator, Date, Language, Publisher, Source).
- Semantics of the resource. The DC.Subject element comprises four
subelements, the academic discipline to be distinguished from the main
concept, its synonyms, and other possible subject terms.
- Pedagogical attributes. These elements designate metadata specific to
the educational character of Ariadne resources.
- user type, either 'learner' or 'author'
- document type, either 'expositive' (learning from instruction or study)
or 'active' (learning by doing)
- document format, from a list of values depending on document type
- usage (optional) freetext comments on how to use the resource
The following elements are applicable to the metadata record only if the
value of user is 'learner', i.e., the resource is for a student
rather than for use by a creator of educational resources
- didactical context, values from a list of learning styles
- course level, a pair of values: country and educational level as
specified in that country (e.g., 'US, K-3')
- difficulty level, 'low', 'medium', or 'high' for designated course
level
- interaction quality (for 'active' resource) semantic density
(for expositive resource), 'low', 'medium', or 'high'
- pedagogical duration, minutes needed by an average learner to use the
resource
- Technical characteristics (document handle for resource retrieval,
format, file size, and installation information)
- Conditions for use (DC.Rights element which can have the values 'free'
or 'not free' and elements for price and acquisition of not free
resources).
- Meta-metadata (creator of the metadata, date, language, and revision
information about the metadata record itself)
Comment: The Ariadne metadata record provides detailed features for
cataloging materials targeted for students, including evaluation components
whose values must be provided by experts. Other materials are not
explicitly provided for: it is not clear how one would catalog information
resources for teachers, administrators, education researchers, etc. who
are neither 'learners' nor 'authors'.
<http://www.imsproject.org/metadata>
- Scope of project: IMS is a cataloging system sponsored by software
corporations, publishers, and institutions of higher learning developing
education resources on the internet.
- Materials being cataloged: Education resources produced by sponsoring
institutions are the priority materials, but the system is designed to be
extensible for a wide variety of resources.
- Metadata elements: An important design feature of IMS metadata is the
definition of distinct sets of metadata elements to characterize different
kinds of resources. Apart from a small base set, there is no attempt to
impose the same metadata on all resources. IMS uses 'containers' to define
the set of metadata appropriate for a particular object (item to be
cataloged).
- The Base Set of metadata contains the minimal elements for description
of
all resources cataloged in IMS. These are basic Dublin Core elements
including
title, publisher, date, description, format, identifier, and subject. The
base
set also includes 'meta-metadata' (author of the metadata, creation date,
date
of last modification, validator, and container type).
- There are two types of values for the Subject element in the Base Set:
the
descriptor, a term from a controlled vocabulary paired with an
identifier naming the source of the the term (e.g., ERIC, LCSH); and key
word, for uncontrolled terms that describe the subject, such as a
proper
nouns and new terminology.
- The container types that augment the Base Set are: Item,
Module, and Tool. The Item container is used for a unitary
resource such as a text or image. The Module container is for a learning
resource with a specific educational value or purpose, such as a course,
topic,
assessment, assignment, or activity. The Tool container is for a learning
resource that provides a function for the user, such as a word processor,
calculator, statistical analysis package, or composition guide.
- The Module container employs the full set of IMS metadata elements; the
Item and Tool containers employ subsets of this set. The Item set adds
only
'author', 'price code' and 'rights' to the Base set, providing minimal
elements
for the description of education resources that are not well-defined
curricular
materials. The Tool set includes the elements 'user support' and
'platform'.
The Module container includes a particularly detailed set of elements for
educational methods and objectives, including objectives mandated by
government
agencies.
- In this product, learning level is indicated by a pair of values
which describe the academic grade and skill level for which the resource is
appropriate. For example, 8-9:3 would be used to represent a resource
appropriate for ages 8 to 9 with a difficulty level of 3 on a scale of
- -5, 5
being most difficult.
- The IMS pedagogy element corresponds to Ariadne's document
type.
It has two possible values: "expository" (= "expositive" in Ariadne)
and
"discovery" (= "active" in Ariadne).
- The IMS resource type is a Dublin Core element, but the
controlled
vocabulary associated with this element is peculiar to IMS.
- The IMS use time element, measured in minutes, corresponds to
pedagogical duration in Ariadne.
Comment: IMS employment of containers with distinct sets of
metadata
elements for different kinds of resources offers the prospect of simplified
cataloging to the extent that unnecessary elements can be excluded from a
record structure. Some IMS values, such as the paired values for learning
level, are not directly meaningful to a searcher and would require
cataloging
expertise to use properly. For more efficient searches, it seems desirable
to
map the searcher's familiar vocabulary (such as "grade level" in the US) to
IMS's neutral values of age in years.
<http://gem.syr.edu/Workbench/index.html
>
- Scope of project: GEM is a US Department of Education-sponsored project
which has developed a metadata structure and cataloging program for
distributed cataloging of K-12 resources.
- Materials being cataloged: The project targeted lesson plans and other
curricular materials as the highest priority for cataloging, as is
reflected in the metadata element set.
- Metadata elements: GEM uses all the Dublin Core elements. The GEM
controlled vocabulary for the Dublin Core Subject element is well developed
for curricular materials, providing two levels of classification, one for
general academic subject area, and a second for specific topic. The ERIC
controlled vocabulary can be designated for non-curricular subject values.
A subelement of the Subject element is available for uncontrolled key
words. The Dublin Core is extended with a variety of elements for K-12
curricular resources:
- An Audience element added to the Dublin Core is further
qualified to designate both the immediate user of the resource (Tool For)
and the student population to be served (Beneficiary). Each of these
subelements has its own GEM controlled vocabulary.
- A Grade element is qualified to designate the K-12 grade level
of the beneficiary or, in a Level subelement, beneficiaries outside
the K-12 range.
- A Pedagogy element has three subelements: (1) teaching
(instructional method); (2) grouping (of students in classrooms); and (3)
assessment.
- Quality and Standard elements are available to represent
evaluations by outside agencies.
- A Duration element corresponds to use time in IMS and
pedagogical duration in Ariadne.
- An elaborate set of qualifiers for the Relation element is used
to associate the many evaluative components of a curricular resource (e.g.,
isRevisionHistory, isContentRating, isPeer Review). Reference to these
components in a full description of the resource may be valuable once a
particular resource has been selected for consideration, but it is not
clear what role these components would play in initial resource discovery.
Comment: Although the full set of GEM elements are designed
primarily for curricular materials, it is possible to identify an
appropriate subset that can be used to catalogue non-curricular materials,
such as those targeted by WALDO. A new controlled vocabulary needs to be
designed for each such element. The GEM architecture allows for addition
of new controlled vocabularies by means of the Scheme qualifier.
GEM comes closer to addressing the needs of non-specialist catalogers than
do the other systems reviewed, with element names and definitions providing
a fairly close match to the likely interests of searchers.
Materials
on sites of concern to the WALDO project include both curricular and
non-curricular resources. Curricular resources are, for the most part,
provided through clearinghouses (such as ENC) or other organizations that
create records for databases to support adoption decisions and inventory
and
ordering needs in addition to basic resource discovery. The components of
these
records that correspond to GEM's set of metadata elements can be mapped
fairly
directly to the GEM record without an additional cataloging procedure,
though
terminology for values of elements like Audience and Resource Type may not
match the GEM scheme.
Non-curricular resources such as reports, training guidelines, directories
and
reference lists, interactive forums and listservs, etc., are not provided
through agencies with cataloging capability and do not require extensive or
detailed bibliographic description. Catalog records for such resources are
critical primarily to improve access to the information they contain. The
WALDO
prototype of the element set for such resources includes the 15 Dublin Core
elements, though only 7 are mandatory:
- Date (of metadata record creation)
- Identifier
- Format
- Language
- Resource Type
- Title
- Subject
- Publisher (Online Provider)
Values for other Dublin Core elements may not be readily obtainable from
the
resource "in hand" or may be less critical to initial discovery of relevant
resources. These elements are optional, but should be used if the values
are
given in the resource being cataloged.
- Description (recommended)
- Creator
- Coverage
- Relation
- Source
- Contributor
- Rights
Additional elements are proposed for a catalog record to support the
discovery
of relevant resources for the education community:
- Audience (mandatory)
- Grade/Education Level (optional)
- Essential Resource (optional)
The elements proposed for cataloging WALDO materials correspond to a
subset of
the elements developed by the GEM project. The two projects are concerned
with
materials addressing similar populations and educational issues. It seems
appropriate to use a container architecture like that of the IMS product,
making the elements of the WALDO prototype available in a module of GEM for
non-curricular materials where the full set of GEM elements is available
for
curricular resources.
The two projects differ in the use of subelements to refine particular
elements
(especially Relation) and in the vocabularies associated with
identification of
Subject, Audience, and Resource Type. To the extent that these
vocabularies
describe different conceptual domains, the Scheme qualifier can be used to
specify different vocabulary authorities. For example, a curricular
resource
would use the GEM vocabulary of K-12 academic disciplines. This information
would be represented as follows (using the conventions for embedding in
HTML
documents with the META tag):
<META NAME="DC.Subject.Level1" SCHEME="GEM" CONTENT="arts">
<META NAME="DC.Subject.Level2" SCHEME="GEM" CONTENT="music">
A non-curricular resource dealing with funding for music instruction in
elementary grades would be cataloged with subject terms selected from the
ERIC
thesaurus or another suitable authority list:
<META NAME="DC.Subject" SCHEME="ERIC" CONTENT="music education; fund
raising; philanthropic foundations; ...">
In other cases the differences in vocabulary may need to be resolved by
seeking
consensus among the broader community of users and producers of education
resources. See discussion below.
Most
projects are substantially in agreement about the critical elements needed
for
education materials, such as 'academic discipline' and 'user level' (i.e.,
material for students must indicate grade or learning level, other
materials
may be targeted to teachers, administrators, parents, etc.). There are
unnecessary differences from one project to another, however, in the
terminology of element labels and in the refinement of elements with
qualifiers. Other problematic differences involve smaller-scale elements
for
details such as price code, standards mapping, or software requirements.
Such
elements of a full biblographic record may be useful once a relevant
resource
has been located but are not likely to play a crucial role in the initial
search and make cataloging more difficult for nonspecialists. Much of
this
detail might be more useful in a separate database under an entry for a
particular document, thus keeping the core scheme simple. Alternatively,
the
mechanism of separate "containers" could be employed to isolate metadata
for
simple resource discovery from elements of a fuller record.
It is likely that these projects will work towards greater
interoperability as
they participate in the ongoing development of the Dublin Core. The recent
memorandum of understanding between IMS, Ariadne, and GEM is a particularly
hopeful development. A viable search procedure for educational documents
will
best be created through consensus about how to derive the necessary
components
from a selection of the existing categories that have required so much
effort
and expert knowledge to devise.
Controlled vocabularies are critical to any metadata system because
diversity
in catalogers' choices of terms reduces retrieval effectiveness. Serious
discrepancies have already arisen among education metadata projects with
respect to the lists of terms that specify values for a given metadata
element.
Although all projects have elements for resource type, subject, and
audience,
no two projects use the same list of terms for the values assigned to any
one
of these elements.
Vocabulary lists for the content values of resource types and subjects need
further development. The Waldo prototype has explored the usefulness of
the
thesaurus of ERIC descriptors as a controlled vocabulary of subject terms,
but
much additional work will be necessary to facilitate subject
identification by
non-specialists. For example, interfaces for catalogers should be
developed
that make it easier to locate existing terms by linking synonyms to a
relatively small number of standard vocabulary items. This procedure
differs
significantly from the one employed in the ERIC thesaurus, which presents
catalogers with a multiplicity of narrow alternatives to a more general
term.
Problems that arise in local development of controlled vocabularies are
illustrated below by comparison of lists for resource types in the Dublin
Core,
EdNA, Ariadne, IMS and GEM.
The Dublin Core Resource Type working group has recommended a set of
primary
values for the Resource Type element. This set of terms classifies
resources
according to the nature of the medium embodying the resource and bears a
close
resemblance to the general material descriptions of standard library
catalogs
[AACR2]:
- text
- image
- sound
- data
- software
- interactive
- physical object
A draft 'structuralist version' of resource types subcategorizes these
genres
with a set of more specific terms in an effort to characterize the resource
more fully. The subtypes of "text," for example, are:
- abstract
- advertisement
- article
- correspondence
- dictionary
- form
- homepage
- index
- manual
- manuscript (i.e., unpublished text not described elsewhere)
- minutes
- monograph
- pamphlet
- poem
- preprint
- proceedings
- promotion
- serial
- tech report
- thesis
Some of these terms are counterintuitive. The term "advertisement," for
example, refers to announcements of job openings, while "promotion" refers
to
what American English speakers would think of as advertisements. There is
no
resource type "announcement," though this is a very common type of Web
publication. The recommended procedure for including locally determined
subtypes not provided in this list is to precede them with an 'x-', e.g.,
'text.x-announcement'.
Developers of the major education metadata systems have each created an
idiosyncratic vocabulary of resource types. Comparison of vocabularies for
resource type reveals wide disparities in the interpretation of this
element
despite the fact that all four systems are designed for cataloging of
education
resources. Inspection of these vocabularies (presented below in
alphabetical
lists) will also make it clear that they bear little relation to the Dublin
Core list.
- course offering
- curriculum
- event
- forum
- individual (i.e., home page)
- links
- message
- organization - educational services
- organization - parent
- organization - professional
- project - curriculum
- project - collaborative
- project - research
- project - students
- project - teachers
- report
- school - primary (home page)
- school - secondary (home page)
- university (home page)
- expositive (e.g., hypertext, video)
- active (e.g., exercise, questionnaire, simulation)
- advertisement
- assessment
- base (undifferentiated)
- collection
- dataset
- document
- example
- exercise
- media resource
- message
- miscellaneous
- reference
- schedule
- simulation
- tool
- tutorial
- activity
- artifact
- catalog record
- community (e.g., listservs, online forums)
- course
- curriculum
- data set
- environment
- form
- lesson plan
- primary source
- project
- realia
- reference
- research study
- service
- tool
- unit of instruction
Comment: It is not clear that the categories listed above would be
useful to educators conducting searches on the web. Educators might well
be
more interested in categories that refer to the function of an item rather
than
to the medium or form relevant for archival purposes. Familiar categories
such
as regulation, policy, guidelines, or news, and newer categories like
directory
(list of links) and listserv are essential. Such functional categories are
missing from some or all of the education lists above. Developers should
review lists used in other education metadata projects to achieve a more
consistent set of categories. The list of publication types associated
with
the ERIC thesaurus is a good source for many of the functional categories
relevant to searches by educators, though attention must be paid to
creating a
set of standard terms for resources unique to the Web.
A
metadata record format that facilitates web searches for educational
materials
and cataloging of such materials by non-specialists can be created from
existing technologies and categories of classification. The Dublin Core
allows
for creation of a manageable set of appropriate classification elements,
which
can be expanded as necessary by appropriate qualifiers. Established
authority
lists such as the ERIC thesaurus, which is actively maintained to reflect
developing educational usage, provide a valuable source of terminology for
controlled vocabularies. But the lack of consistency in selection of
vocabularies as values of specific elements poses a substantial problem for
metadata creators and searchers.
This problem of terminological control cannot be resolved without active
collaboration among producers and users of metadata. One alternative to
having
all terms in a rich but unwieldy authority list like the ERIC thesaurus
would
be to separate out terms into categories that can serve as values of
discrete
elements. Instead of having publication types, population groups,
pedagogical
techniques, student groupings, etc. in a single list of subject terms,
these
components could be available as values of specific elements that are
associated only with appropriate resources.
Another strategy might be to create broad conceptual classes of synonyms
and
related terms to which a term entered by a cataloger (or a user in a
search)
would be assigned automatically.
In order to accomplish the task of cataloging and providing access to
present
and future Web resources it is critical that metadata should be both
simple to
create and simple to use. Elements of resource description that are not
critical to resource discovery should be isolated from the critical set of
elements. For WALDO materials, the set of elements required to describe
curricular materials should be distinguished from the smaller set that can
characterize non-curricular resources quickly and efficiently. This can be
accomplished within the GEM cataloging system by using a container
architecture
like that of the IMS to maintain distinct element sets.
[AACR2] Anglo-American Cataloguing Rules, 2d edition, 1988 rev.
[Bibliography] Digital Libraries: Metadata Resources.
http://ifla.inist.fr/ifla/II/metadata.htm
[CSI] US Department of Education, Office of Educational Research and
Improvement, Cross-Site Indexing
<http://165.224.220.67:8765/csi/>
[DC 1] Weibel, Stuart, Jean Godby, Eric Miller, Ron Daniel. OCLC/NCSA
Metadata
Workshop Report, 1995.
http://www.oclc.org:5047/oclc/research/conferences/metadata/dublin_core_report
[DC 2] Dempsey, Lorcan and Stuart L. Weibel. The Warwick Metadata
Workshop: A
Framework for the Deployment of Resource Description. D-Lib Magazine, July
1996. http://www.dlib.org/dlib/july96/07weibel.html
[DC-4] Weibel, Stuart, Renato Iannella, and Warwick Cathro. The Fourth
Dublin
Core Metadata Workshop Report. D-Lib Magazine, June 1997.
http://www.dlib.org/dlib/june97/metadata/06weibel.html
[DC-5] Weibel, Stuart. DC-5: The Helsinki Metadata Workshop. D-Lib
Magazine,
February 1998.
http://www.dlib.org/dlib/february98/metadata/02weibel.html
[DC-REF] Weibel, Stuart and Eric Miller. Dublin Core Metadata Element Set:
Reference Description. 1997
http://purl.org/metadata/dublin_core_elements
[ENC] Eisenhower National Clearinghouse Resource Finder
http://www.enc.org/rf/index.htm
[ERIC] ERIC Database
http://www.aspensys.com/eric/searchdb/dbchart.html
[H&D] Heery, Rachel, and Lorcan Dempsey. "A Review of Metadata: a
survey of
current resource description formats." 1996.
http://www.ukoln.ac.uk/metadata/desire/overview
[HRST] Hearst, Marti A. "Interfaces for Searching the Web," Scientific
American, March 1997, 68-72.
[IR] Sparck-Jones, Karen and Peter Willett. Readings in Information
Retrieval.
Morgan Kaufman, 1997.
[LAB] Northeast and Islands Regional Educational Laboratory at Brown
University, cross-lab index.
<http://www.lab.brown.edu/public/index.shtml>
[NMP] Hakala, Juha, Preben Hansen, Ole Husby, Traugott Koch, and Susanne
Thorborg. The Nordic metadata project, final report. 1998.
http://linnea.helsinki.fi/meta/nmfinal.htm
[RFC1] Weibel, Stuart L., John A Kunze, and Carl Lagoze. Dublin Core
Metadata
for Simple Resource Discovery, 1998.
ftp://ietf.org/internet-drafts/draft-kunze-dc-02.txt
[RUS] Russom, Jacqueline. Metadata for Information Retrieval on the
Internet:
Background for WALDO Project.
[SAL] Salton, Gerard. Automatic Text Processing. Addison-Wesley, 1989.
|
