We'll have an hour of short talks on search and various search
engines, followed by more informal discussion and hacking. It will be
a perfect time to add search to your app, try out a new search engine
and ask questions. We'll have a Ruby focus, but plenty to interest
I will present:
"An Introduction to ElasticSearch"
Pat Allen, creator of Thinking Sphinx and Flying Sphinx, will discuss:
"Full-text search with Sphinx"
Florian Hanke, creator of the Picky search engine, will give us:
"A peek into search engines / Picky: Ruby, Fast, Flexible (pick three)"
And Tal Rotbart, founder of SpringSense, will give a lightening talk on:
"Meaning Based Search with Solr"
6pm for a 6:30 start
Monday 29th August
Level 1, 41 Stewart St
Richmond, VIC 3121
Call: 0403 899 534 if you have any questions on the night.
RSVP: not required, but appreciated.
Thanks Chris, the search engine hack night sounds interesting and I hope I can
attend. A good chance to refresh my rusty Ruby skillz too.
So yeah I'll like to see how Picky works and the other talk look good also.
I remember writing a simple play-toy fulltext search thing in Perl
once apon a time, I was surprised that it was reasonably fast.
One search problem I'd like to know about is how to do pattern search in a
nodes-and-arcs graph. e.g. if you had a graph representing some math equation,
searching for a structure like a*(b+c) and replacing it with a*b + a*c.
This could be part of a bigger 'simplify' or 'solve' search.
Also could search for structures in a complex molecule, etc.
Hopefully it could have something like the regexp repetition operators like ? +
* and other useful stuff.
There must be some comp-sci / graph theory stuff out there on how to do this
efficently. I haven't googled for it yet :)
I was also interested in a 'people powered' web search engine, not sure how
practical that would be.
Well I was thinking to use instant messaging. Someone types in a query
maybe "good cookie recipe with spice but no nuts".
The query goes to a 'general handler' person who is online, not a specialist.
They would redirect it to a cooking expert, who might give the answer or
redirect it to a pastry chef / baking specialist. That person can give a
good response, in the form of a single suggestion, a link to a list of recipes
on some webpage or whatever. A diolog could take place at any stage,
to clarify the query or just for fun :)
I think it would work best if at each stage two or more helpers are contacted,
in case one is not there, or to get a range of responses. Or this might be a
search option. A random or 'round robin' system would be used. If there are
100 cooks online, it might send the query to 1 or 2 of them.
Of course this might only work well once lots of people are helping,
however the number of helpers would I suppose grow with the number of users,
and without specialists, an expert searcher or an ordinary geek like myself
could probably find something decent on a regular search engine for them.
The response time would be much slower than e.g. google, maybe from 10 to 30
seconds or more, but the quality of the response could be much better.
And it could be more fun perhaps, since you are interacting with real people.
I think there are some similar search systems with 'human cogs' out there on
the web, but the ones I looked at looked overly commercial for my liking.
I'm not sure if I want to dedicate a substantial chunk of my life to sitting in
front of the computer helping other people find things, but anyway that was the
idea. As I spend a lot of time at the computer anyway, it might be okay.