On face value, the questions driving research into the next generation of search engines may appear simple enough — “What exactly is this person looking for and how do we guide them to the best information about it” — but the work involved is mind-bogglingly complex and goes to the very fabric of the web as we know it.

Search engines are readily confused. A common example is the keyword “Paris” — does this person mean Hilton, or the one in France, or maybe Texas?. But in every query, and especially as people shift to use more “natural language” in their searches, engines such as Google have to sort through levels of ambiguity. A good way to solve that problem, it’s agreed, is for all objects — places, people, songs, even individual genes — to have a “key”, a kind of unique identifier observed the world over. (In the jargon, this is known as a “universal resource indicator”.) Each time a web page made reference to that object, the code of the page would reference the key. That way engines would know precisely to which object a page is referring. This “semantic” approach to search, as its known, would achieve two things. First it would enable Google to understand the content of web pages with greater certainty, but more importantly, it would allow search engines to extract specific bits of information out of different web pages and serve them up as one “result”, rather than the way it works now, where an engine lists many pages, each of which may be relevant in its own way.

Say, for instance, two scientific papers are published online — separately — which make reference to a particular gene. At the moment, Google has no way of “knowing” whether the two papers are talking about the same gene. If the gene had a key, it could. Search engines would thus be able to draw much more sophisticated connections between information on different web pages.

One challenge — among the many — is how you decide what key will be used to refer to the billions of objects that may be tagged in this way, and that requires some deep “under the bonnet” maintenance on the web, not to mention international consensus — both of which are still very much works in progress.

Bing

Microsoft’s new search offering. The quality of results has improved greatly from its previous incarnation, Live.com. In look and feel, quite similar to Google, but on the results page, there’s a thin vertical line at the end of each result, which, when you click it, shows a more in-depth preview of the content on the results page — helpful for gauging the usefulness of searches. A stream of pictures also rotates behind the home page.

Ask

Ask has a nice “page preview” feature: when you click on a pair of binoculars next to a result, you’re shown a snapshot of the page — a helpful way of steering clear of ugly, cluttered pages. The Q&A tab also tries to pre-empt full sentence questions you may have in relation to your keyword. For instance, if you type “gladiators”, the Q&A tab will give answers (from Yahoo! Answers) to questions such as “Why did gladiators fight?”

Google

Try clicking “Show options” on the left of a results page to get richer results. You can select “Past 24 hours” or “Past week” to display more recent content, or elect always to see images from results pages as well as text. “Wonderwheel”, also under Show options, is a quirky graphical way of exploring topics similar to the one you looked for. A search for “cheese”, for instance, quickly takes you to the related “Cheese Maturing Webcam”, and to cheddarvision.tv — where you can watch cheese ripening in the West Country.

Explore Tech & Web

• Personal Tech

• The Web

• Gadgets & Gaming