Introduction

All too often in our professional roles as knowledge workers we encounter online information that is incorrect. The memorable incidents tend to be the more outrageously incorrect, but there are other forms of error that while more minor are far more frequent, and hinder our ability to perform our information work.

A memorable (and ultimately motivational) example for me was the experience I had a while back visiting the local public library with my daughter. In attempting to lookup in the on-line Web catalogue what books in the Asterix the Gaul series by René Goscinny and Albert Uderzo the library had, I became concerned that the author search was not returning all the matches it should, as there were titles missing from the result set that I knew I'd previously had out on loan. Further investigation revealed the cause—in over 70% of the cases the author's name René Goscinny were wrong! Having tracked down the books through other means in the system, variants of the name that appeared included Ren? Goscinny, Renþ Goscinny, and even Ren☐ Goscinny. I kid you not!

Working with my collegues Dave Nichols and Mike Twidale, we have come up with a Web browser extension that assists users when they encounter such errors, and in particular those pesky but all too frequent legitimate variants of peoples names that crop up when looking up information in digital library systems: with and without accents, with and without middle names, and so on, that confound searching.

The system works by letting the user manipulate, merge and/or expand any terms appearing live within in a web page. Through this sequence a user might, for example, change Renþ to René, and then merge this item with the existing René on the page. Clicking on this merge item then brings up all the search results for both versions of the name. Built-in transform exists for commonly occuring situations, such as removing middle names, and standardizing punctuation of initials, but essential the user is free to change the text any way they see fit.

We've called the system Computer Says No ... Computer Says Maybe ... Computer Says Yes, or CSN for short. With a nod to the fictional character Carol Beer from the comedy show Little Britain who has the catch phrase "Computer says no" the name was chosen to reflect the fact that the approach provides a way for users to go beyond the frustrating—from their point of view—read-only nature of the web pages produced by today's digital library systems, and actually change the content of pages to something more correct and meaningful when problems arise.

Download

Computer Says No ... Maybe ... Yes—which it must be pointed out is more of a proof-of-concept than a production system—is available for download through the following link:

The script was developed as a GreaseMonkey user-script for Firefox, and is known to also work in Chrome, Internet Explorer, Opera, and Safari but the set up can be a bit more involved.

Quick Start

If you are interested in trying out CSN with the minimum of fuss, we recommend you install GreaseMonkey in Firefox (if you do not already have it installed), restart the web browser and then click on the above link to download the CSN script. This will trigger GreaseMonkey's mechanism for installing user-scripts.

The Gory Details (aka the not so quick start)

More detailed instructions of how to install CSN in a variety of Web browsers can be found here.

The Next Step

With the install complete, you're now ready to take CSN for a test drive (but do please remember it is a proof-of-concept rather than a production system). If you'd like a more detailed description of how to use CSN, I suggest you have a read of the example walk-through.

I hope you find the extension useful.

David Bainbridge
Department of Computer Science
University of Waikato
New Zealand