Lexxe is a search engine which aims to provide answers (and links) when you ask a question by processing mainly unstructured texts using natural language technology. It claims to be a third-generation search engine as it focuses on linguistics over the symbolic logic of second-generation search engines like Google.

The idea seems to be semantic and somewhat Web 3.0; it processes the question to try to understand it and then processes results to try to find an answer. Also, just to quickly look at its coding the site looks like its built on ColdFusion.

Naturally, the machine’s judgement is far from perfect. If I type in What is my name? it returns Earl Tv, obviously understanding My Name is Earl. Still the fact it can like What is my name? with My name is… is quite an achievement. However, this would allow abuse of the system much like Googlebombing. Somethings are obviously just wrong: Who is a failure? and the answer is Congestive Heart.

The site also provides links to related pages, as well as clusters of similar keywords. This, to be honest, isn’t very useful. For the time being Google is much better at returning links, and can cluster words together using Google Sets.

Lexxe is an interest attempt at artificial search technology, though it doesn’t work well. Still, early days. Perhaps in 10 years we’ll have Web 3.0 and sites like this will prosper. Then again, we may not and Google may continue to.

Tags: web2.0, web2, web 2.0, web 2, web 3, web3.0, web 3.0, web3, semantic web, search engine, lexxe