Merlot Federated Search
This was an important topic for me as we’ve been looking into a wildly federated solution to try and accomplish a single portal to access distance ed in BC. So what was interesting to me was to compare that problem with this one for similarities and differences.
I guess mostly differences – in this model (quite rightly) they don’t worry about lack of response from any of the federated catalogues – if, for whatever reason, the catalogue’s not responding in time for one particular query session, it is just left out of those results. This seems fair enough, though the decision on MERLOT’s part not to indicate this lack of responsiveness back to the user seems strange. Their answer was that it was confusing to the user, but it seems to me a little ‘Catalogues not responding to this query’ link or list is simple enough.
They’ve made a bunch of decisions that seem pretty sensible:
- limit responses to 25 per service, partly to limit harvesting and partly I assume to assure better overall responsiveness (and probably only that many are needed in any case);
- model their service partly around the google API (I think I’m getting this right) which I assume is in part because this is one people are now becoming acquainted with and thus can repurpose interfaces/learning;
- offer an intermediary page between query and results to deal with response-time waits, keep user informed of status.
They seemed to think that network latencies weren’t a problem (one of the first federated partners is in Australia) and that this would scale very easily, which againbecause of the lack of requirement that all catalogues MUST respond is probably quite true.
All and all seemed quite promising – SWL