If you really want to know what a semantic search capability is, you can either read what I’ve written here or simply push on the “cloud button” in the right hand column. I’ve not only been briefed on this technology, I’ve actually implemented it. That’s sounds a little grandiose, but implementing it didn’t involve much more work than adding a video to a posting, because it amounts to the same thing. You paste a little bit of javascript onto a page.

What do you mean, Semantic?

The burgeoning hype around Web 3.0 is almost palpable. It’s certainly surfable. The underlying reality is simply this:

Searching with current search engines is truly limited.

It is word-based, not meaning-based. You search the web using only “atoms” of meaning (i.e. single words) that are completely devoid of context. To provide a simple example, if I search for “queen” the search engine can only guess whether I’m interested in female monarchs, or universities, colleges, ships or hotels that bear the name Queen or the rock band called Queen.

As search engine users we try to get around this by searching on several words, but while this may help with context, it has the unwelcome consequence of excluding some web pages we might be interested in. So if we search on “Queen Rock Band” we will improve context significantly – although we may still retrieve some pages that tell of a rock band playing in the presence of a female monarch, or at some college or university that bears the name “Queen.” We will also eliminate some web pages about Queen, the Rock Band that just don’t happen to mention the words “rock band” on the page. That’s the tyranny of the exact match.

Sponsored Link: Semantic Targeting

The great hope of the semantic web (Web 3.0) is that we will be able to search on meaning rather than words and as we search we will not inadvertently exclude pages we may have an interest in, and we will definitely exclude the irrelevant. It will be like using a rifle rather than a blunderbuss.

Quintura as a Semantic Search Capability

Quintura provides a cloud of words, like a tag cloud, (just like del.icio.us and many blogs) but the tag cloud is not generated by people, it is determined according to a kind of tree of meaning (you could use the term “ontology”). Quintura indexes a site in order to build its semantic map of the pages. This results in a set of indexes that is significantly better than a tag cloud.

You can step through the cloud and it will retrieve relevant web pages according to the path you take. Alternatively you can add one term from the cloud to another and it will narrow the search, but it will do so according to meaning. You can enter any word and it will form a “cloud of meaning” around that word based on the pages it has indexed.

It’s best to play with Quintura to appreciate what it can do. I found it fun and also useful, and that’s why I implemented it. There are about 750 pages on this web site, amounting to somewhere between half a million and a million words. Quintura has indexed them all and, it has linked together their “meaning” – so you can now search this site semantically. There’s also a Google Search on this site, so you can compare the results that the two searches give you. (I concluded that Quintura was significantly better, but I already know what’s on this site ,so I don’t correspond to a typical user.)

Quintura has one brilliant capability in that, as you step from one word to another in its “cloud of words” it throws up relevant pages dynamically. The experience is much better than descending through page after page of irrelevant results.

When you get bored with going through this web site, you can go to Quintura.com and surf the whole web.

This kind of search capability will become commonplace soon. It’s a genuine improvement.