Full Text Indexing versus Tagging
I gave a talk about “Full Text Indexing versus Tagging, or how to tag full text index” at webtuesday. The slides of the talk can be found here.
Information and Links
Join the fray by commenting, tracking what others have to say, or linking to it from your blog.
Reader Comments
Dear Balaji, the thing that can’t be seen from the slides and you are missing was a demonstration of a simple test application while giving the talk which generated these two graphs. The setup was as following: I had a lucene index containing a subset of all wikipedia pages and a set of 10′000 frequent but random terms. These terms were used as queries and also as tags. I then started one application which was asking the queries (red line in graph, shows number of queries answered per second) and a few seconds later started another application which tagged one random wikipedia page out of the index with a random term by adding a new term to a seperate field (green line, shows the number of tags applied per second).
The first graph shows the behavior when you take lucene and just follow a standard description on how to query and make updates to the index (e.g. according to “lucene in action”). The second graph shows the same numbers if you use an optimized setup for this, i.e. the setup shown (in a simplified way) on page 36. The experiment ran on a laptop and there was no intention to bring the absolute numbers down, just to show the order of difference for the two approaches. There was no DB involved in this setup, lucene for everything.
Does that answer your question?
Thanks for commenting, Felix
Dear Felix,
Came across your presentation pdf
on various approaches to tagging.
One thing i did not understand was the charts. You have two charts and a blank page. Are these charts for DB for everyting, combination of DB and lucene , lucene for everything?
What do these charts represent