Search spelt differentlyPosted: February 5, 2007
I have taken the time to put together what I’d consider an Alpha version of the search, semantic and NLP (natural language processing) techniques I’ve been talking about to give an idea of the power that’s going to be available to us within the year most likely.
Initially I’ve set up a very simple example of how a search engine (any search engine in fact could use the technique I’ve employed here) can interpret the web pages in its index to pick up the nuances of a word and those related to it. It’ll pick up current affairs (for those of you in the uk, try entering police – the word peerages will most likely appear) and whilst talking to a friend about it over a few beers the other night; I commented on the fact that it appears to have it’s own opinion if you enter superlatives or negatives as a search term.He quickly realised my error and corrected me: “…it doesn’t have it’s own opinion, it has the average opinion of all the web pages in its index…”.
For now it’s a fun tool, as ever on this blog though I’m asking you to think of the possibilities available, to kick start you here are a few:
- Self aware text, every single word on your page tagged (idea stolen from evolvingtrends).
- Self aware pages, automatically have your meta tags and content written with a Surrch API.
- A favorite of mine: personal search engines, you index the topics you want by specifying one word for each, the semantic layer takes care of the rest and finds the related keywords for you.
The website currently runs on a beleagured pentium4 on a limited upload line, with the continually expanding search indexes sitting on the same box. It may not be fast but hopefully will allow you to garner some interesting thoughts.
As always, let me know if you have any thoughts in the comments and apologies for the lack of writing but this programming lark keeps me extremely busy. The first demo will be up later this week if all goes to plan (server freebies and support welcome!) at http://www.surrch.com.