Reading about the recently released
Pew search engine use report (
via LibrarianInBlack) [
pdf] only serves to remind us all how searching has become an inherent part of online life.
Search is a really important thing to me. Almost everything that I do is focused around making material accessible to the end user (either now or over time) and search is a huge part of that. Search isn't just something that the public use, it's also of critical importance to staff, but it's the end user who is likely to have the most difficulty with a new search interface.
When I was a researcher, and a user of the resources I now play a part in providing, I sought information in ways which seem somewhat insane to me now. How I ever found anything of use is beyond me. Let me give you a quick précis of my "search strategy":
- Fire a keyword into the OPAC (at this point I was only looking at printed material, I had no idea that there was anything else available or how to search it)
- Scan down the results and order anything that looked vaguely relevant (it's free right?)1
- Look at my stack of material and immediately sort it based on the cover, blurb and index
- Read the four remaining books and note down the references that looked useful
- Go back to the OPAC and request those items (or at least those that I could easily find). Rinse and repeat.
(For me, journals weren't something you searched; you'd read the last 5 years worth and only go further back if there were specific articles referenced in other papers.)
That little vignette helps to remind me that even though I now know about authorities, subject headings, cross references and formats there's no reason that end users do or even should.
The perennial question with searching surrounds how much effort you expect users to put in to the locating of material. The problem is that every searcher, every potential user, has a different level of
attention2 that they're willing to spend on the search process. Here's a few examples:
An undergraduate student might be happy with any resource which gives them some level of information and might spend very little time searching. These are the searchers who might favour instant access (think e-Journals) and small results sets.
A family history researcher might be looking for one specific document (a probate record, a historic map, etc) or for any information related to a particular individual. They might have higher reserves of attention but they are looking for specific items, reference number searches, combinations and authorities are key for these people.
A professional researcher (academic or otherwise) might want to be more comprehensive, they might want access to a wide range of unique material or to take long and complex journeys through policy documents. Their level of attention is high when it comes to exploring the material but lower when it comes to locating that material. Unlike the undergraduate they're not looking for 'a resource' but 'the resource' however in contrast to the family history researcher they might not have a clear idea what that resource is or even what form it might take. For these types of searchers the ability to easily combine complex searches is key.
Members of staff are likely to be looking for specific records rather than items (as cataloguers or desk staff) and have the highest level of attention. They can tap into the resources required to use back-end search tools and to restrict searches by elements which they can be confident are in the data and they have the knowledge of the collections required to limit searches from the outset. Staff will know to use control numbers where possible and to follow the rules of controlled vocabulary.
Of course, in a perfect world we could answer all these needs simply by presenting the user with an empty search box, google-style. However, we have to remember that google and libraries aren't the same thing. Firstly, google's success and revenue comes from accumulating small amounts of attention from a large amount of people - in essence, it's in google's interests to provide the resource that's selected only at the point where the users attention is almost exhausted (of course, you can't bank on any specific level of attention so it's best to try and get the most appropriate result first).
Secondly, google indexes data more than metadata. Despite the trends towards the digital most of our material as cultural organisations is not searchable in any kind of thorough fashion. Instead we're limited to hand made metadata records which are not only subjective but also tread the line between brevity and usefulness. Finally, the creators of web content (and the systems which serve that content) are often concerned with making it easy for google to index their data, if publishers of traditional material had the same incentive perhaps we'd see them providing high quality catalogue records themselves?
How do we give the reader the tools they need to find the resources they want within the attention they're willing to spend? If we think they'll read them and make use of them we can produce guides, give training sessions and create FAQs and this kind of strategy will work for members of staff.
But for users these kind of approaches aren't ever going to be as effective. A struggling searcher might look to help guides and FAQs if they are confident that the item they seek is in the collections, or might make an enquiry about the item at the desk or via email. Or, they might just as easily give up. Realistically it's too much to expect that users will pre-read documentation on searching before just entering a search but we shouldn't under estimate the willingness of users to learn from failed searches and to improve their own strategies (assuming that the data they're searching is consistent enough to support this...).
I can see how facets,
FRBR/RDA,
WorldCat, exposure to google and other record combining approaches will help to move us beyond the current library OPAC search provision but even then I'm not convinced that they represent a panacea. It's interesting to note that some libraries are now providing a request form for material that cannot be found on the OPAC (note: that's subtly different from material that isn't on the OPAC and accepts that the gap between some users and the OPAC is too large).
We often try and elevate google as a shining example of simple searching for complex data but we have to remember that they're starting from a different position. If google were to swap content with a large library or archive it's probable that they would start their resource discovery work with a programme of mass digitisation which would facilitate the full indexing of the resources. Of course, that's one option for libraries but in lieu of a huge input of funds we can still take steps to improve search-based resource discovery.
- Improve metadata - by making use of auxiliary resources, upgrading records from other systems and by making sure that our cataloguing is not only consistent but also focuses on the provision of metadata hooks which explicitly facilitate search strategies which we expect users to use.
- Improve Search Interfaces - if the metadata is the fuel, the search mechanism is the engine. It's not unreasonable to expect that search interfaces will need to be tweaked, upgraded and even replaced on a more regular cycle than library systems have required in the past.
- Expose Your Metadata - Every hit on your metadata, regardless of where that takes place, is a potential use. Exposing good quality metadata to sector and non-sector resources (including search engines) is key to getting someone into the system. If someone only uses google to find records in your library system this does not constitute a failure - they've found the resource haven't they?
- Provide Simple Help Focused On Particular Uses - Where there's attention to be spent provide resources which shortcut the learning process (including making good use of the "zero results" page). However, general information is only useful if you're looking to support general searching - if you know your users want to search one particular collection provide a very simple 'how-to' for that collection with some tips.
- Study Users - Ouch. It had to come didn't it? Usability studies cost money and are often ineffective. A large proportion of studies are conducted with pre-determined groups (often existing users) and flag up detailed problems which experienced find irritating - this can be useful but you need to think about all those people who spent their attention long before becoming an experienced user and as a result feel alienated.
1 This is why I ordered one particular children's book, with an appropriate seeming title, over and over again. Subject headings? What are subject headings?2 If you haven't read Davenport & Beck's book you need to - seriously.
