lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Leo Galambos <Le...@seznam.cz>
Subject Another engine
Date Mon, 09 Sep 2002 23:35:51 GMT
Hi.

I like Lucene (THAT'S RIGHT!), but it doesn't offer me all features I 
want. That's why I decided to write another JAVA engine. If the features 
(see below) are interested for you, and you are a developer, that would 
like to help me with the new engine, PLEASE let me know (use my private 
mail, I DO NOT WANT TO START A FLAMEWAR HERE, LARBIN IS COOL. Howgh). 
Thank you.

I would like to contribute to Lucene project, but I have chosen 
different object model for the new engine... :-(

Demo runs here: http://somis4.ais.dundee.ac.uk/sheeef/index.jsp (the 
machine indexes *.ac.uk right now, so the speed may be slower if you try 
many concurrent queries).

Features:
- extended Boolean model with p-metrics
- index compression via Golomb, Elias-Gamma, and block coding. Better 
than Lucene for more than 20-50%. Each inverted list is stored in the 
best coding method. The method is selected by "inverted list metadata" 
object - it is not hard-coded.
- highly configurable dynamization algorithm - it guarantees a good 
response time for query(), insert(), delete() operations (without 
degradation of index structure)
- universal stemming technique for almost any language (not used in demo)
- on distributed architecture, insert() would not lock the index
- the engine would be able to simulate Harvest structure of Brokers
- ...

Speed (indexing 2000 HTML documents, without stemming)
Larbin-latest: 1'17"
the engine: 1'22"
[RH73,IBMJDK131+JIT]

Regards,

Leo




--
To unsubscribe, e-mail:   <mailto:lucene-dev-unsubscribe@jakarta.apache.org>
For additional commands, e-mail: <mailto:lucene-dev-help@jakarta.apache.org>


Mime
View raw message