Web Services

November 17th has now passed, and so we’ve begun migrating new search boxes on to the UCL site. These new searches work with our  Google Search Appliances, and allow us to customise and refine UCL search results.

One advantage of using the GSAs, which we didn’t mention in the previous post, is that they can give us far more information about how visitors use our search boxes. For instance, we were surprised to find these terms being the most popular search queries after we first began to migrate the new search boxes:

Where were department names, news items, or members of staff? Generally, these were far lower in the list. It seems that the majority of searches at UCL were queries relating to specific IT services – particularly email. More amusingly, the top 100 queries also featured ‘squirel mail’, ‘squirell mail’, ‘squirelmail’ and a good number of other misspellings of the webmail system. Still more people are searching for these terms than almost anything else in the university, and assuming they’ve spelt ‘squirrel’ wrong, they’re not going to receive helpful results.

Thankfully however, the GSA allows us to add key matches to such queries (a bit like the sponsored links on the main Google searches), or simply arrange results listings to give more useful information for these quirks. There’s still lots to learn about how best to arrange the search engine results, and with users searching for over 50,000 unique queries in the last week alone, this could be a massive job. But if it helps users consistently get the information in as efficient a time as possible, it could be a genuinely useful tool for many in UCL.

In the last few weeks Web Services and Information Systems have been busy setting up two new Google Search Appliances (GSA) to be used throughout the UCL website. A GSA is a server that uses Google’s search technology, but is under UCL’s control: we can refine search results to better suit our own requirements.

For some UCL sites, the new GSAs could help immensely. For instance, when a user currently searches for e.g. ‘Silva’, they’ll have to trawl through a number of other listings to find anything about the Silva Content Management System. Because the Web Services Silva pages don’t have such a high Google ranking score (as eg. they may not have been linked to from external pages), they’ll appear lower in the listings. And sadly, this problem occurs with many other Google search terms: in many cases, the pages that users are searching for are simply not those at the top of the result listings (or anywhere near, in some cases).

So when GSA is implemented here, UCL will be able to take full control: We can refine results according to certain search queries; and we can add extra promotion for departmental homepages and key documents. Ultimately, we’ll be able to ensure that sensible results will always occur (and if they don’t at first, at least we can amend the system so that they eventually do!).

GSA also allows us to:

• search not only the public content of UCL domains,  but also (using authentication) search restricted content.
• search other areas like document management systems and collaboration tools (although these will be introduced progressively)

From November 17th, we’ll begin migrating search boxes on UCL web pages from the general Google search to the new GSA.