Search log potluck
We (Lou and Rich) have been able to poke into a few sample logs from a variety of organizations. But for privacy reasons it's been difficult to come by samples that could be publicly shared. It's a pity, as some of you have already asked to see log samples. And it'd be useful to compare formats provided by different search engine vendors.
So how about we provide the venue (right here on our "book in progress" site) and you provide the raw samples? We don't need entire files; just enough (say, a couple dozen queries) so their respective syntaxes are clear and understandable. As an added bonus, we'll be glad to mask the IP addresses to protect searcher's privacy.
A small library of publicly-accessible search log samples would be useful to so many of us in so many ways; please give us a hand and send to lou (at) louisrosenfeld.com the following:
- A dozen or two lines of your logs
- The name of the search engine that generated the log
- If you have it, an explanation of each of the log's components (i.e., which part is the query, which is the number of results)
- A URL for the site that generated these queries
- Your name and title
We'll be sure to acknowledge your help; thanks!
Log samples:
Comments
Did I ever give you guys some of our log files?
Posted by: K.G. Schneider | July 10, 2006 10:27 PM
No, please send them!
Posted by: Lou Rosenfeld | July 12, 2006 10:55 AM
This is a homegrown log format that we use on the Digital Learning Commons website (http://www.learningcommons.org/). The search engine is really three search engines (hence the "type" field in the log): a federated search of library databases, a search of our online courses, and a basic website search. Due to some technical limitations about how all of these pieces are stitched together, the federated search engine doesn't return hit counts to this log (we do track that elsewhere, though), hence the zeros for many of the queries.
We also have a little homegrown reporting system...basically tells us the most recent queries and the most popular over a given amount of time. The latter is useful when populating our best bets system (also homegrown).
Let me know if you have questions about this. (And the search engine itself is password protected...email me at karln@learningcommons.org if you'd like a demo password.)
Karl
term datestamp type web_hits catalog_hits
basque 20060712165432 courses 0 0
visual art benefit 20060712182256 library 0 0
visual art brain 20060712182436 site 17 20
visual art brain 20060712182506 site 17 20
login 20060712195655 library 0 0
catalyst 20060712215039 library 0 0
Economics 20060712224720 site 40 15
school for the blind 20060713074628 site 248 84
washington state school for the blind 20060713074753 site 248 305
"state school for the blind" 20060713074817 library 0 0
state school for the blind 20060713074836 site 248 288
the necklace 20060713084545 library 0 0
george washington 20060713111213 library 0 0
mexico 20060713120228 library 0 0
European Union 20060713172345 library 0 0
french 20060713172348 library 0 0
teamwork 20060713212530 library 0 0
our friend judith summary 20060714094734 library 0 0
Layers of the earth 20060714095452 library 0 0
Posted by: Karl Nelson | July 14, 2006 03:39 PM
Perhaps if you contact the vendors themselves? They should be somewhat forthcoming..
Posted by: Peter | July 17, 2006 08:46 AM
Peter: You'd think so.
Karl: Thanks!!
Posted by: Lou Rosenfeld | July 17, 2006 10:50 AM
I can provide some logs. To what address should I send them? How many do you want?
Posted by: Robert Blakeley | July 18, 2006 03:52 PM