quantifying repository success

The current D-Lib Magazine issue has an article written by Leslie Carr and Tim Brody of the University of Southampton on repository assessment.

The authors describe the creation of an assessment system that can be used by "external observers." From a systems standpoint, this is supported through OAI-PMH harvesting and data housed in Celestial, an OAI-PMH harvesting proxy. (Footnote 2 of the article suggests that OAI-PMH data, by itself, couldn't be used to accurately track repository submission patterns. Thus, the external observers are also saving and manipulating data.) Also, as the authors note, the code of the DSpace repository software helps to drive submission patterns, in its canned support of one, two, or three peer review, workflow steps and the possibility of batch submission by a final reviewer (though they argue that this fact shouldn't fundamentaly alter repository assessement).

The value of sustained deposit is completely comprehensible as an indicator of "faculty-repository engagement." The question is: could repositories characterized by large, infrequent batch loads be considered a success? By the commonly-referenced Clifford Lynch definition that's cited in the article, I would argue that the answer is yes. Obviously, sustained deposit is more desirable and indicates a greater level of faculty buy-in to the repository service. If, however, most of the materials are submitted by a small number of intermediaries, infrequent and heavy deposit activity is more understandable.

Carr and Brody clearly appreciate the complexity of repository metrics. Additionally, the article informs readers about ROAR and the data that it provides on repository submission patterns. They also identify IRs (see Table 2) that have relatively high rates of sustained deposit.