‘Real-time’ search engine makes Web change direction
There are plenty of Internet firms competing to produce “real-time” searching in which the time it takes between content going online and appearing in search results is cut to as short as possible. But one site that’s just begun public testing aims to achieve this goal by reversing the flow of data.
Collecta takes advantage of a technology known as Extensible Messaging and Presence Protocol (XMPP). The key to XMPP is that it alters the way one site gets information from others. In traditional Web site set-ups, a computer pulls the information from the original source (which is why when you load a Web page, it’s known technically as a “request”).
XMPP works on a push, rather than pull, basis meaning that as soon as information appears online, it’s automatically sent to any other computers which have requested it. In this case, anything published on the WordPress blogging system automatically goes to Collecta’s system. That means, according to the firm, that a new blog post can appear in its results in just half a second.
However, only a few sites use XMPP at the moment, so Collecta pads out its “real-time” results by constantly trawling sites which have frequent updates such as Twitter and YouTube.
When I first used the site, I assumed there was some sort of bug as nothing appeared to be happening. It turns out that’s an inherent drawback to the site: unless something’s been posted momentarily before your search, you won’t get any results until a relevant post is actually made somewhere. Unless you choose a particularly popular term, you could be staring at a blank screen for minutes on end.
While Collecta is an interesting technological idea, it’s going to be a tough sell. Even the site’s own management say only about one in five Web searches have any element of timeliness to them that might make it worthwhile. And there’s not yet any ranking of results in relevance, importance or credibility: it’s simply a barrage of matching terms.
Another ‘real-time’ search site which has also just launched in beta, crowdeye, takes a more old-school approach (or as much as a Twitter-search can be old-school). Rather than searching the entire Web, it’s simply searches Twitter posts then presents the results along with some popular relevant Web pages, stats on how popular the topic has been recently, and some suggested associated terms to search for.
Related Posts:
