A Tale of Two Prototypes

This page has been set up to demonstrate two prototypes that use OpenSearch to provide access to the UK Web Archive. Instead of using something like:

in your web browser for searching (located top-right in Firefox), once you have install the search extension for the UK Web Archive, you can start to use:

The first prototype is pretty much what you would expect: setting your web browser to using the UKWA Search box (set from a pull-down menu in Firefox) and typing search terms into it followed by pressing enter or clicking on the magnifying glass icon initiates a search at the UK Web Archive, and the main page to the user's browser loads up with the results (see the section below for details on how to install it).

The second prototype:

adds a twist to the former prototype. Given that a strong motivator for someone accessing the UK Web Archive is because something that previously had existed on-line is no longer there, or else the person is interested in how the content of a page/site has changed over time, this got me to thinking that it would be quite nice to support user searching where the result of a query shows the result of both searching the web at large as it currently is and from the UK Web Archive at the same time.

To this end, this is what the second prototype does. If you look at the image snapshot except above, you'll notice that the instructive text that appears in the search box (and disappears immediately the user clicks in the box) says, "UK Dual Web Search". This is because the result of initiating a search in this version is to show the results from searching the UK Web Archive and Google (restricted to only pages in the .uk domain) side-by-side:

(Click on figure to access full size version.)

The above image is a snapshot of using the second prototype to search for the term JISC, which gives a flavour of how this plays out, but why not install the prototypes—covered in the next sectoin—and try them out for yourself?

Installing the prototypes

I've mentioned that these are prototypes, right? What this equates to is that I've only tested things in Firefox—but there is no reason to think they wouldn't work in any of the other popular browsers, just that what you have to do to install it will probably be a bit different.

Being a prototype, I haven't gone as far as listing it on any of the browser add-on sites. Instead, to install it (in Firefox anyway) you can make use of the auto-discovery feature of OpenSearch. The page you are viewing declares the existence of the two prototype search engine quick-search boxes. Clicking on the drop-down menu icon on the left-hand side of your Firefox's quick-search box brings up a list of other installed search engine to choose from, and below that the option to add the two new ones this page declares. In my installation of Firefox it looks like:

and a click on each "Add ..." line gets them installed. All going well, you should now be in a position to try the two prototypes for yourself.

If you'd like to try setting the two prototypes in a different web browser, here are the raw application/opensearchdescription+xml files:

You'll probably want to download them locally, and then add them in to your browser (somehow!).


Obviously the prototypes are very rough and ready (I think I've spent more time writing this web page than coding!) but they give an indication of what is possible using this web browser standard. For simplicity the dual search restricts to .uk domains to deliver a Google search that is at least comparable to the sites indexed by the UK Web Archive. The first technical improvement would be to change this so it matches exactly the sites indexed by UKWA. And notwithstanding the overwhelming dominance of this company as the search engine people use (Andy, was it over 85% you had on one of your slides at the Digital Research conference?) another improvement would be to let the user determine which search engine was used in the comparison with UKWA.

Thinking further ahead for the dual search version, to support more generally the intended use of locating "what's changed?" and "what's now missing?" its functionally would need to be expanded. For the what has changed type of search, it would probably make more sense to have the live web search (i.e., the Google query) take the lead, and for each item returned then, in the side-by-side view presented to the user, provide the corresponding URL link in the UKWA (in a Wayback Machine kind of way) to give the user easy access to see how the page has changed.

To develop support for users searching for information that is no longer there, this intriguing possibility would be a more significant undertaking, and (I suspect) require a deeper level of integration through the planned APIs.

I hope you find what is provided here useful, or at least food for throught.

David Bainbridge,
email: [email protected]