Metasearch – Everybody knows

A report on a Metasearch survey among RLG Members was posted on the RLG website the other day. The survey was done during May and June and consisted of interviews with a small set of institutions that are at various stages of implementing metasearch facilities for their communities. I’m not sure that this topic has the same interest across libraries, museums and archives although the materials in all these institutions are targets for the meta of all metasearch applications – web search engines. It’s really libraries who seem most driven to add value to the large array of information resources that they license by putting a metasearch engine on top. From the survey it’s clear that one of the goals is to provide a simplified user interface where results will be merged. There were a number of things that surprised me – the relative indifference to effective ranking algorithms, the extent to which this was viewed as a tool to help undergraduates get started on research, and the disdain for efforts like Google Scholar that seem to have already established the search paradigm for the undergraduate audience. The survey was confirmatory about a number of directions to which RLG has already committed – designing Archival Resources as a target for metasearch engines is a low priority because of this predicted undergraduate focus and we’re putting more effort into making RLG image resources interoperate with other image aggregations rather than worrying about how they behave as a target. We wondered whether what we gleaned from these interviews matches well with what you see at your institution. Tell us.

We’ve already gotten a very thoughtful response from Roy Tennant at the California Digital Library that I’ve posted with his permission, as a comment.

5 Comments on “Metasearch – Everybody knows”

  1. I’m a bit confused by all this. Maybe it’s the long word: “metasearch”? I’ve certainly never liked the meta-web-search engines, but I love union catalogues, and tools like OAIster. The union catalogue is just a particular implementation of metasearch. Isn’t Google just a way of meta-searching a few billion web-sites?

    In the UK, JISC has been looking at this area in various guises for around ten years, first with the idea of the DNER (Distributed National Electronic Resource) and now with the Information Environment. For reasons that frustrate the hell out of me, this hasn’t got anywhere much. But there is some progress; see the EDINA service Getref for example

    My current University library, in common with most others, offers hundreds or even thousands of databases (just counted 78 in the A to E part of the alphabetic index, just of the online section). I work with the JISC Geospatial Working Group, aiming to find and negotiate for useful datasets in that area. One of the problems is getting libraries to buy and users to use these datasets. Sometimes this is about money, often it’s about the economy of attention. People just don’t register it’s there; you told them when they weren’t interested, and when they would have needed it, they don’t remember, and don’t see it in the huge lists. Metasearch can help in this, bringing reslts to your attention that might prompt you to try that resource.

    I can see many ways in which metasearch in various forms would be useful to all levels of academia. Given the negative result and the concerns that has raised here, maybe a wider survey should be undertaken, perhaps after review of the original instrument (which I have not seen)?


  2. Arnold,

    Unfortunately, I believe it is the latter. That is, I would be much happier about the state of things overall if I felt it was your survey methodology that was at fault. Instead, I fear that most libraries are still at an early state of comprehension of how best to use metasearching. Even as far along this road as I think we are, we realize that there is both a lot we still need to find out and experiment with as well as some serious technical issues standing in our way. As arrogant as it may sound, I think we have a more mature sense of how best to position these services within a large research library environment. It’s my belief that large research libraries will eventually come around to this view as well, so long as they haven’t thrown the baby out with the bathwater by then.


  3. Roy,

    Thanks for sending your reactions to our metasearch survey. I was dismayed by what we heard too, or what we thought we heard. What you’ve been doing does sound different–much more deliberate and focused. I’d agree that that would be the way to create a valuable alternative to Google Scholar. I was dismayed because we didn’t hear our respondents talk about that kind of focus, but about a general-purpose tool for undergraduates.

    You say, very graciously, that the trouble is not with our survey, but with the early state of system implementations. I wonder, though, whether you think our way of choosing a sample or asking our questions or interpreting the responses might have misrepresented the general situation. Do you think we’ve read the general state of these metasearch efforts wrong? Or that they’re mostly not yet at the point where they need to be? You can imagine how interested we’d be in your answer to this question.


  4. I found your Metasearch Survey both interesting and somewhat troubling. Troubling, since I think it exhibits a rather early and perhaps naive view of both the possibilities and problems of metasearching. This isn’t the fault of your survey, but of the early state of system implementations within libraries.

    To be more specific, I was concerned to see items like the following in your report, since I think they are based on unproven suppositions and conjecture (my comments on each follow in brackets):

    – “faculty too may need ways to discover what licensed resources are available to them; however, that wasn’t ordinarily viewed as a purpose of metasearch” [ we see metasearch software as a platform by which we can integrate access to many resources and services — not just licensed databases; therefore, one of the modules we seek to develop is a database recommender service, which would be used within a metasearch portal to highlight additional databases of interest]

    – “few respondents mentioned deduplication as an important feature of a metasearch product” [ this is particularly amazing to me given the fact that most thought metasearching was for one-stop searching for undergraduates — this is clearly the audience/purpose for which deduplication is most important]

    – “Most respondents had organized their metasearch efforts around disciplines…they reported this was not because they saw this as preferable, but because they saw it as expedient; the relevant sets of resources were already identified.” [ this is completely counter to our model of created search interfaces tailored to specific audiences and/or needs – we are all about trying to understand a particular need and crafting the right metasearch service for that need]

    I would be loathe to make any far-reaching decisions on such an early set of impressions, especially when based on a small sample of institutions with very little metasearching experience. At CDL, we have had a metasearch service deployed for over five years. We have had a great deal of time to think about what works and doesn’t work in a large research library environment, and we are building on that experience with needs assessment activities where we speak with undergraduates, graduate students, and faculty about their specific research needs (these reports are available from our project web site.)

    Overall, here at CDL we have a very different (and I believe both more realistic and more robust) vision of library-based metasearching. We see metasearch software as merely an integrative infrastructure that will allow us to stitch together easy and effective access to licensed database content, remote repositories that surface their metadata for harvesting, and web sites that can be crawled and indexed. We envision a database recommender tool that will begin by suggesting other databases to search and that will eventually allow us to dynamically create a metasearch tailored to a specific query. We seek, in other words, a rich infrastructure that enables librarians to craft rich and yet easy to use and effective search services on behalf of a certain clientele with particular needs. If we are successful, large, generic search services such as Google Scholar will pale in comparison. Thanks,
    Roy Tennant
    CDL Metasearch Infrastructure Project

Comments are closed.