The ACM Digital Libary is an invaluable research resource for a wide range of academic disciplines. I would find it hard to write papers without it. Invaluable as it is, it could be more valuable still by supporting not just use by its subscribers, but also research by its subscribers.

Several SIGs (SIGIR, SIGWEB and SIGCHI come to mind) represent research areas related to information retrieval, digital libraries, and user interface design. Researchers in these areas often use collections of documents (e.g., TREC) to run experiments to test improvements in algorithms and interfaces. While this can be a fruitful approach to achieve incremental progress, some of the research results get trapped in experimental systems and are not made available to users for a variety of reasons. Cost is certainly one reason, but also the lack of collections that people actually *want* or *need* to search in their daily lives is another big impediment.

So here is where the ACM DL can be even more useful: if the ACM were to make its collection available for research use by its subscribers, the research results obtained by them could be used to inform the design of the digital library. There are several possible models for such collaboration, but probably the least expensive (for ACM) is to provide a means of mirroring the collection on some non-ACM site for research purposes. Updates to the mirror could be applied periodically, but the two collections need not be completely synchronized.

By providing such a research resource, the ACM could foster a range of research whose results could be transferred directly. Usability testing could be done on moderately large scale without compromising the integrity of the existing system. New indexing or retrieval algorithms could be applied and compared with the existing baseline. Behavioral studies of exploratory search could be conducted in a naturalistic way.

There are clearly reasons for doing nothing, given the current economic climate and ACM’s budgetary constraints. But much of the cost of supporting this research infrastructure could be pushed out to the research organizations interested in using the collections, without affecting ACM’s hosting and serving load. Issues of intellectual property could be addressed as they are today, by granting ip-address-based site licenses or individual accounts. As the number of users would not change, the load on the authentication component (which could still be managed by ACM) would remain relatively unchanged.

I encourage interested ACM members to take up this call, to make our valuable research resource a valuable research tool as well.

Share on: