Why Aquabrowser isn’t doing it for me

Aquabrowser is very tasty looking, but I’m not convinced — as I pointed out in Ten brainless things an online academic library can do (Aquabrowser was number 10). Note that I wrote academic library and not just plain-old library; the reasons for this is that I have experience working at academic libraries, and know the specific problems we face, but I would guess that the same things apply to other types of library too.

This article is based on an admittedly brief assessment of Aquabrowser — I have done quite a lot of different searches, but these searches have been selected to represent what I think is typical for the platform. The assessment is based on use of implementations on three different sites, one non-academic, two academic. The findings presented here cannot to be taken as definitive, but I believe them to be indicative of Aquabrowser’s suitability as a discovery platform for academic libraries.

If you’re not sure what Aquabrowser is, take a look at any of the sites below, or a look at the Aquabrowser Library page on the Aquabrowser website.

The Aquabrowser implementations I looked at can be found at the following sites:

• Queen’s library — this is the “demo site” recommended as the demo on the Aquabrowser website
• Oklahoma State’s Big Orange Search System
• The University of Chicago’s Lens

I chose these sites, because they were a) sites advertised on the Aquabrwser web site, and sites that had been mentioned to me. I hope that they are representative!

What I did was perform a series of searches for the following terms, click the links to review the searches:

• polysemy [Queen’s][Oklahoma][Chicago]
• minimalism grammar [Queen’s][Oklahoma][Chicago]
• aphasia [Queen’s][Oklahoma][Chicago]

Note: I only used simple search, because I found that Advanced search actually provided very narrow results sets that didn’t necessarily provide any advantage over simple OPACs.

Seach: Polysemy

• all systems returned highly relevant results immediately; Q(ueen’s): 4, O(klahoma State): 40; C(hicago): 160
• the discovery panes were all equally lacklustre.
  • Lens managed to produce a single thesaurus term “semantics”, and some “association” terms that were odd; clicking these association terms proved a mixed bag: metaphor produced some good results, but synonym proved to be of little help (though the first two results were in a language I don’t speak)
• Many of the cluster facets presented by the systems are useful; these would allow a student writing a term paper to narrow the search down to relevant literature, though on face value, Queen’s facets were more effective, because there was much less “noise” (they have smaller holdings of this kind of literature)

Search: minimalism grammar

• Again, all systems produced good results immediately; Q: 11, O:87, C: 240
• Discover panes were odd again, some good associations, but a lot of weak ones:
  

  Term type	Queens	Oklahoma	Chicago
  Spelling variants	acquisition, hebrew, comparative, english, arabic, linguistics, semantics, review, essential, russian, focus, language, general, reference, outline, context	Grimmer, grammars, grammer, minimalist, maximalism	Grimmer, grammer, gramar, minimalist, Grammmar
  Associations	acquisition, hebrew, comparative, english, arabic, linguistics, semantics, review, essential, russian, focus, language, general, reference, outline, context	schoole, outline, ornament, logic, russian, mauger, claudius, review, introduction, language, composition, dictionary, reference, transformational, latine, shorte, english, reader, hebrew	teaching, french, outline, reader, principles, note, greek, new, hebrew, school, latin, essay, general, language, testament, key, russian, introduction, compleat
  Translations	gramatica, gramática, Sprachlehre Grammatik	gramatica	gramatica

• Cluster facets were again most useful for narrowing down the current results set

Search: Aphasia

In this search, I got a taste of what the discover pane in Aquabrowser does well: taking general, common terms to link to more specific concepts the user to discover things about the library holdings.

• All systems returned good results immediately; Q:18, O:254, C: 492
  • The University of Chicago system expanded my search and told me which ones with half a page of terms that made it impossible to see the first results without scrolling
• The discover panes provided good links:
  

  Term type	Queens	Oklahoma	Chicago
  Spelling variants	Awassa, Apasia, Aphakia	Dysphasia, Aphasias, Biphasic, aphakia, aphasic	Aphasias, aphaia, aphakia, ephesia
  Associations	neurolinguistics, speech, status, aphasic, brookshire, nell, assessment, communicative	Coursebook, acquire, disorder, rehabilitation, neurogenic, therapy, adult, apraxia, agnosia, intervention, fluent, kindred	intellect, dyslexia, phonetics, childhood, agnosia, neuropsychology, recovery, neurolinguistics, aphasic, psycholoinguistics, apraxia, cerebrovascular, minnesota, damage, acquire
  Translations	n/a	afasia	afasia
  Thesaurus	n/a	n/a	language disorders, Landau-Klettner Syndrome, Brain Diseases, Dominance, Cerebral, Aphasia, Acquired

• Cluster facets provide the usual level of helpfulness

General remarks

Where Aquabrowser was really strong was on simple searches; try searching for a really very general term like “psychology”, and the discover pane of Aquabrowser really shines. On the other hand, this makes the catalogue a knowledge discovery platform of a kind that is quite foreign, and thereby probably counterintuitive for the majority of users (when did you last input “psychology” in your OPAC/Google and expect quality results?) Of course, people will “get” this system after quarter of an hour of interaction, but do people actually use the discover pane? I wouldn’t, but perhaps I’m not representative.

The engine behind this part of Aquabrowser does not seem to be particularly intelligent in that entering results in any OPAC is likely to return similar results in the first instance; in the second instance — clicking the discovery terms, simple boolean, combinatory searches will fail to impress librarians, or users of Google. Real intelligence would filter results, make proper associations and present proper thesaurus terms from a controlled vocabulary — potentially with an explanation of these.

Discovery graphs are largely a waste of time unless you’re doing very basic searching where you really don’t know what you’re searching for. Because the the graphs reflect the indexing used in the OPAC, these can be very weak; there is obviously little or no real intelligence behind this element. Mis-spelling in spelling suggestions is a particular minus (grammmar at the University of Chicago).

The clustering facets are work well, but this is also simple, taking two simple terms that recur in two or more records, and then ranking these according to hit frequency. I could bodge together a system that did this very badly and slowly for BIBSYS Ask in fifteen minutes — it would be too slow to be usable because everything would depend on network transport, but the point would be made.

The fact that the three columns don’t seem always to interact makes Aquabrowser a real loser in my eyes; why do the discover and refine panes not change when a results item is clicked, why does the discover pane not change when a refine item is clicked?

I also experienced a lot of frustrating session problems and timeouts on both Queens and BOSS. There also seemed to be issues using browser navigation, all in all, the user experience was a bit lacking. The technology behind Aquabrowser is ASP, so this is obviously a problem that isn’t going to go away — we’re back to brainless things (ASP is thing number 12, but I’ve written about it before). Another huge minus is that Aquabrowser has typically been installed in a frame that makes copying a URL more difficult than it should be; hiding the path to the business end of the search does not make the search page RESTful!

My evaluation is therefore that Aquabrowser is prettier than a typical OPAC, but it in reality provides little more than a thin layer of prettiness, leaving all the ugliness of OPACs fully in place. What I see as lacking is “intelligence” in the system, and as such, this is still necessary at the end-user end if you want to use Aquabrowser, or indeed any OPAC, successfully. My advice: save your money and train people how to use your system as it is. I still like the cluster facets though.

Tags: academic libraries, Aquabrowser, knowledge discovery, OPAC