Wednesday, 19 March 2008

Boolean is like Schrodinger's cat...

Notes from the Web workshop (February 15, 2008, at McGill University, Montreal) session called "Is Boolean Dead? Research and the New Web" 1 hour presentation by Rajiv Johal and Beth Dunning, Librarians, Howard Ross Library of Management, McGill University.

Having driven from Kingston and not accounting enough for the traffic in Montreal, I was late for this presentation by a few minutes so I missed the opening statements but was flung fully into the meat of the presentation without any preamble.

The question was asked, "What is Search 2.0?" There is no clear definition for this concept but it does include the increasingly popular use of facets and tags, and is emphatically not Boolean. Some examples of public search engines illustrated this 2.0 idea are:
  • the child focussed, visually appealing search engine RedZee [],
  • the guided search engine ChaCha [], and
  • Rollyo [] which allows you to choose or create a tool that searches only a certain set of sites.
The open source ILS movement is also including some very 2.0 concepts:
  • Fish4info [] is an example of a user-friendly interface for your OPAC/ILS created with the CMS Drupal; and
  • Evergreen [], which is quite popular in the US, is a full OPAC/ILS product with facets, reviews, and no Boolean.
In many of these tools, although Boolean operators are not used or not recognized in the default or basic searches, very often there is still the option to be able to use it. It just remains hidden to the user who may be confused or ignorant about Boolean techniques.

Increasingly, vendor solutions are also integrating some of the Search 2.0 preferences, whether as full products or “middleware”, pieces you can simply add to your current system. And example is Aquabrowser’s tag cloud (note: the tag cloud is frequently on the left of the screen because research has shown that that is where average users tend to look first), used at the University of Chicago’s library catalog option “Lens” []. When trying to determine whether this technology was a useful addition to their catalog, the U of C found that upwards of 9 out of 10 students (even Masters and PhD students) improve their search results with access to the tag cloud. (I’m not sure that the survey really shows that it is a good tool: it could just be that generally people are really bad searchers and that any tool would improve things. Searching problems are very often a matter of lack of a clear process not always lack of synonyms or proper terms.) U of C is still maintaining their original catalog interface (sans tag cloud) so that all the usual functionality can still be used included Boolean searching.

At this point, an audience member asked about about the existence of tools that enabled virtual browsing of the shelf but the presenters were not aware of any such tools.

Citation and other research databases were also entering into the Search 2.0 ring. For example Factiva Search 2.0 ( a business news database) does not translate “and” or “or” into the Boolean functions but rather includes them as any other keyword in the search because AND or OR could be a company’s ticker code or part of one. When searching using a term that the tool recognizes as a company’s ticker, it breaks this option out to link to the company’s homepage or related news, and it includes a multimedia search in the platform as well.

Another question from the audience concerned the existence of recommender-type tools in use in any of these tools. Apparently McMaster University Library is testing this functionality.

Another vendor jumping on the Search 2.0 bandwagon is EBSCO with their Visual Search. This uses indexing to create a visual interface to explore articles but, just like the other tools, it really only helps with browsing as opposed to performing comprehensive searches.

The important points to take away from this session were summarized as:
  • Boolean searching is not dead but rather hidden from the average user; and
  • Search 2.0 is here to stay, being about what users want. Try implementing some of the ideas in some small way.
After-session comments from the audience:
  • Consider a search tool used by Walmart called Endeca [].
  • What would you do first? Remove the necessity to search with Boolean operators and include a spell check feature. The term “fuzzy searching” was used to describe these kinds of functions.
  • Consider the tension between teaching users how to search and making it easier for the users to search.

No comments: