Channel #semsol: Logs

This is a public chat log generated from the #semsol IRC channel.

08:46:10 thosh: hey guys
08:46:12 thosh: there is a small bug in the ARC2 zip from 2008-08-28:
08:46:13 thosh: some parentheses missing in ARC2_RemoteStore.php on line 132:
08:46:14 thosh: switch $qt {
10:54:26 bengee: cheers, thosh, I'll fix it
10:59:21 bengee: ok, fixed, I've (lazily) just replaced the zip
14:56:39 thosh: maybe it's OT here - but i have a beginner's question re. querying dbpedia
14:56:41 thosh: i am working on a small script that suggests dbpedia uris for user submitted strings
14:56:42 thosh: i guess it is inevitable that such a sparql query takes quite some time? like this one:
14:56:44 thosh: SELECT ?entity ?abstract WHERE {
14:56:45 thosh: ?entity dbpedia2:abstract ?abstract .
14:56:47 thosh: ?entity rdfs:label ?label .
14:56:48 thosh: FILTER regex(str(?label), "^Ludwig Wittgenstein") .
14:56:50 thosh: FILTER (lang(?abstract )='de') .
14:56:51 thosh: }
14:57:14 thosh: it takes about a minute to answer that query. any chance of bringing that down to <10s?
15:00:01 bengee: hmm, is it faster w/o the lang() filter maybe?
15:00:23 bengee: I found dbpedia quite fast for basic regex queries
15:00:51 bengee: there might also be a magic property for FULLTEXT queries
15:02:29 bengee: ?label bif:contains "Ludwig Wittgenstein*" might work
15:02:55 thosh: hm, without the filter i immediatly get an error on the web based UI: "The estimated execution time 286 (sec) exceeds the limit of 240 (sec)"
15:03:20 bengee: did you try adding a LIMIT
15:03:23 bengee: ?
15:04:45 thosh: bengee: aha, i haven't heard of that fulltext queriesor bif:contains
15:05:21 bengee: it's a proprietary extension of the sparql engine powering dbpedia
15:06:19 thosh: thx, i will look at that
15:06:57 thosh: just tried limit, it helps insofar that it prevents the above error message
15:08:12 thosh: when i add another pattern like "?entity dbpedia2:birthPlace <http://dbpedia.org/resource/Vienna> ." the regex is really quick.
15:27:18 thosh: bif:contains does help a lot, thanks bengee!
15:37:35 bengee: cool
15:39:59 thosh: though dbpedia is now down :-p
15:40:23 bengee: oops
15:43:07 MacTed: (I'm looking into the down dbpedia...)
16:01:26 mmmmmrob: mmmmmrob waves to bengee
16:01:35 bengee: heya
16:01:44 mmmmmrob: bengee: have you had any thoughts about caching in ARC2?
16:02:04 mmmmmrob: mmmmmrob realises he should have checked if it does any before asking
16:02:13 bengee: there is none yet
16:03:16 mmmmmrob: have you had any design ideas/thoughts? memcached, disk caches, how to invalidate etc?
16:03:24 bengee: current plan was to implement that on a higher level (i.e. in the Trice framework I'm addin on top of ARC), but I've started considering caching at the ARC-level
16:03:59 mmmmmrob: bengee: ok, I'm just prototyping some at the app level to, maybe we should comapre notes later?
16:05:03 bengee: the approach I#ve taken in Trice so far is a separate ARC store for caching, where an entry has an ID, a timestamp, a path/category, and the data
16:06:00 bengee: so you can check for a certain snippet directly, or via timestamp/expiration, and you can delete cached entries with a certain path/category
16:06:39 mmmmmrob: sounds interesting
16:06:42 MacTed: (dbpedia should be working now ... sorry for the glitch)
16:06:51 mmmmmrob: trice isn't avalable to look at yet?
16:07:06 bengee: still not, no :(
16:07:33 mmmmmrob: bengee: that's ok, don't worry :-)
16:10:27 thosh: MacTed: hm, it did work some minutes ago, but i can't reach it now...
16:11:15 MacTed: hm. checking. (I'm not in the office today, just happened to note your mention that it was down, so am relaying...)
16:19:21 MacTed: arc-bot, pointer?
16:19:30 MacTed: arc-bot, poiner
16:19:33 MacTed: arc-bot, pointer
16:19:44 bengee: stupid bot ;)
16:20:05 MacTed: my typing didn't help
16:20:31 MacTed: OK, dbpedia *should* be back now. upgrading server had a hiccup.
16:24:29 thosh: MacTed: yep, it's here, great
16:27:16 MacTed: cool. please ping me if it goes down again!
16:32:02 thosh: MacTed: yeah, unfortuanetly it is gone again
16:32:19 MacTed: one of those days, clearly.
16:49:59 MacTed: hrm. thosh is gone, but dbpedia should be back again.