Peer Review

Peer Review Report general comments from Stu Weibel:

(Within this section Developer responses are italicised)

This report is the single most comprehensive survey of metadatastandards and issues that I am aware of. It is a very useful resource in a rapidly changing environment that is characterized by a proliferation of communities with different understandings of resource discovery, different requirements for description, different vocabularies for describing similar concepts, and legacy systems that mitigate against convergence of approach and process.

The report provides a comparative morphology, if you will, of many of the emerging formats in the metadata arena. It provides a valuable resource both for those immersed in the technology and for those who are entering the fray and seeking educational information to support their understanding.

The rapidly changing environment of network resource description makes it a particularly valuable resource, but also exposes the problem that anyone trying to enter the field will inevitably have. Many important issues have changed in the month since it was first written: there is a revised Dublin Core set, additional workshops have taken place, network software vendors deploy new systems, the importance of some protocols wanes as others wax (does anyone still use Gopher and WAIS, for example?).

There is no protection from this, but it raises the question of how to manage and document such change, and what the role any static report such as this can be a volatile context. Even as a snapshot in time of the metadata ecology, it is an important resource. My own hope is that this document will become the basis of a dynamically updated clearinghouse of metadata formats. This isn't an ultimate solution, but is certainly an important step along the way. Michael Day's collection of pointers to crosswalks is a very important contribution in this direction as well (and, I suspect, has arisen from precisely thisconcern about rapid obsolescence).

If I may wax philosophic for a moment, I think this work points out an interesting contrast between the thrusts of digital library work in the UK and the US. The US approach has focussed resources on a small number of large projects, the substance of which is more speculative and further out on the horizon. One may expect interesting and innovative results from these projects, but it is hard to see how their results will fit into the information services or educational computing environments of the next 5 years or so.

The UK eLib projects appear to be smaller, more diversified, and more practically focussed. This seems to me to have higher potential for return, certainly as far as the practical issues of resource discovery go. I applaud and encourage efforts such as these. They feed my optimism about the difficult business of fostering progress that is founded as much in sociological change as in scientific or technical advances.

Some minor niggles and nits:

In the section Metadata and its Uses, the discussion on coupling of the metadata record and its referent should be elaborated... this is a critical issue that is widely misunderstood, and the implications of different approaches should be discussed (embedded metadata versus closely coupled versus loosely coupled).

(This will be addressed as part of the ongoing work within the project on resource description. Experimental work is being undertaken at UKOLN to investigate holding metadata separately from its referent and this will be fed back into the working papers of the project.)

The figure at the top of page 8 should be made more clear. It identifies the general categories of the formats discussed and hence provides a framework for organizing the various individual schemes discussed subsequently, but it needs further formalization. What do the axes mean? Left to write presumably suggests simple to complex... is there any significance to the 3-D nature of the boxes? Should each of the three boxes have a label? How are the contents of each related? Much of this is explored in the text, but a diagram of this sort should stand alone (and have a title and caption, as well as clear labels).

(The figure has been revised into tabular format to aid clarity.)

The Dublin Core work has changed substantially from the representation here, as the authors will no doubt know, having participated actively in those changes. My presumption is that other formats might have experienced similar changes. This is not a criticism, but rather an observation in support of the problems identified earlier in this review.

(It is our intention to update and improve access to this work as part of further project work on resource description.)

Others of the formats described in this report are more stable. It might be useful to add in the first section a brief discussion (or table?) of the characteristics of the schemes, to summarize explicitly the classification of the schemes as stable, experimental, formally controlled, open or proprietary...? This information is identified in the discussions of the formats, but some of it might be usefully extracted and summarized in the introductory section on general principles. As a practical matter, many readers will not dig deeply into the meat of the format descriptions, wanting only a brief introduction such as the first section provides.

A summary of general characteristics in the front of the report might be useful (or perhaps even a middle section,

that offers a thumbnail comparison of the schemes in tabular format?

(We will consider elaborating the Typology Table, and/or including more tables to provide this information as part of further project work on resource description.)

Congratulations on a job well done, a service to the community that I hope will be sustained.

Next	Table of Contents