[00:06:36] *** Quits: daherb (~daherb@ppp-93-104-160-26.dynamic.mnet-online.de) (Quit: leaving)
[08:35:33] *** Quits: vin-ivar (~vinit@122.169.38.69) (Remote host closed the connection)
[08:36:09] *** Joins: vin-ivar (~vinit@122.169.38.69)
[09:34:15] *** Quits: spectre (~fran@115.114.202.84.customer.cdi.no) (Ping timeout: 244 seconds)
[10:18:37] *** Joins: spectre (~fran@dhcp507-ans.wifi.uit.no)
[10:22:23] *** Joins: jmvanel (~jmvanel@crz75-3-78-193-235-5.fbxo.proxad.net)
[13:15:17] <inariksit> hello GF class! ^__^
[13:17:24] *** Joins: iwk (~user@flv49020.hum.gu.se)
[13:17:27] <inariksit> \o/
[13:17:38] <inariksit> no ice cream today, but moral ice cream for iwk
[13:17:51] <iwk> awesome! :D
[13:18:05] *** Joins: LukeG (82f13112@gateway/web/freenode/ip.130.241.49.18)
[13:22:02] <inariksit> https://github.com/inariksit/GF/tree/master/lib/src/translator
[13:22:16] <inariksit> sorry not that
[13:22:31] <inariksit> https://github.com/GrammaticalFramework/GF/tree/master/lib/src/translator
[13:22:33] <iwk> "that was my fork, it has some weird things going on, don't do that" <--- QOTD
[13:22:35] <inariksit> ^----- that
[13:23:02] <inariksit> iwk: hahah
[13:23:02] <LukeG> ^_^
[13:23:12] <iwk> ;)
[13:23:24] <inariksit> well the weird things are just in the PGF library, so you're safe :-P
[13:23:26] *** Joins: ehrle (82f13110@gateway/web/freenode/ip.130.241.49.16)
[13:23:31] <inariksit> https://github.com/GrammaticalFramework/GF/tree/master/lib/src/translator
[13:24:31] *** Joins: Liina (82f106c4@gateway/web/freenode/ip.130.241.6.196)
[13:24:51] <inariksit> \o/
[13:24:53] <inariksit> https://github.com/GrammaticalFramework/GF/tree/master/lib/src/translator
[13:25:03] <LukeG> Third time is the charm
[13:25:26] <LukeG> I wonder if that is an idiom in a GF lexicon somewhere
[13:25:38] <iwk> "3 is the magic number"
[13:35:01] *** Quits: Liina (82f106c4@gateway/web/freenode/ip.130.241.6.196) (Ping timeout: 246 seconds)
[13:48:34] <inariksit> ok so let's keep track!
[13:48:56] <inariksit> ehrle: from beginning to exophthalmus
[13:49:00] <inariksit> iwk: from exorbitance to end
[13:49:08] <inariksit> Liina: from beginning to matter_V
[13:49:28] <inariksit> and LukeG: everything, do semantics or e.g. eliminate stupidity
[13:49:40] <inariksit> like british/american spellings in abstract syntax
[13:51:09] <inariksit> now to part 1
[13:51:11] <inariksit> "find examples where GF does clearly better than Google translate, and vice versa. Try to analyse the errors."
[13:51:58] <inariksit> http://cloud.grammaticalframework.org/wc.html
[13:51:59] <inariksit> use this
[13:52:41] <inariksit> if you find something funny, post it here!
[13:55:00] <inariksit> example:
[13:55:05] <inariksit> original sentence "Dom: Svenska staten tar över piratebay.se"
[13:55:13] <inariksit> translation "Sentence: state juniper of Svenska takes over piratebay. To see"
[13:55:50] <iwk> "gnuer går inte att rida på" --- "Gnus do not go to ride on"
[13:56:07] <iwk> (Correct translation: "it is not possible to ride on gnus")
[13:57:06] <iwk> Guess the poor computer can't detect the idiomatic use of "gå"
[13:58:47] <iwk> Word class disambiguation fail: "det ligger i en hög hög" -- "it lies in a high high"
[13:59:35] <iwk> (OK, none of those pertain to the actual task because it was GF that failed, but I couldn't resist :P)
[14:00:30] <iwk> (and google fares no better=
[14:06:40] <LukeG> :D
[14:08:39] <iwk> I accidentally typed "jav" into the GF wide coverage translation demo
[14:08:46] <iwk> I got "yeah-off"
[14:08:50] <iwk> What the heck is that?
[14:09:26] <iwk> Sorry, *yea-off. but whatever
[14:10:34] *** Joins: daherb (~daherb@ppp-93-104-172-166.dynamic.mnet-online.de)
[14:14:30] <inariksit> haha wtf
[14:14:58] <inariksit> sometimes the GF translation does weird word splitting
[14:16:10] <iwk> probably combines "ja" with "av" for some sick reason
[14:16:31] <inariksit> hihi
[14:16:49] <inariksit> I collected once some weirdest things in various NLP tools
[14:16:51] <iwk> Oh, and I've actually found something interesting now:
[14:16:51] <inariksit> let's see if I find it
[14:16:54] <inariksit> cool!
[14:17:06] <iwk> the sentence "jag vill inte ha någon vombat"
[14:17:17] <inariksit> hahah
[14:17:19] <iwk> leaving aside the depressing fact that neither GT nor GF knows what a wombat is
[14:17:23] <inariksit> ;__;
[14:17:37] <iwk> GT translates "I do not want any vombat"
[14:17:43] <inariksit> that's good!
[14:17:49] <inariksit> syntactically at least :-P
[14:17:58] <iwk> While GF gives us "I do not want to have someone vombat"
[14:18:07] <inariksit> aww
[14:18:14] <inariksit> oh shit I only now noticed it was G*T*
[14:18:20] <inariksit> I thought "wow nice GF"
[14:18:30] <inariksit> but yeah, that's the chunking :-P
[14:18:34] <iwk> :'(
[14:18:45] <inariksit> one out of vocabulary item makes it to go for chunking
[14:18:53] <inariksit> so it loses information of the syntactic structure
[14:19:06] <iwk> OK. Sensible fall-back, of course, so we shouldn't be too harsh
[14:19:11] <inariksit> yeah :)
[14:19:35] <iwk> but illustrates an area where statistics outperform rules (shhh!)
[14:19:36] <inariksit> one of my favourite GF translations is "Hello, doubtless yarmulke an experience it slopes be anglicized xmas it lived devoid move it pays to go of your ti."
[14:19:41] <inariksit> yeah, definitely
[14:19:45] <inariksit> also lexical selection
[14:19:47] <iwk> Oh that's priceless :D
[14:19:47] *** Joins: lt_ (82f106c4@gateway/web/freenode/ip.130.241.6.196)
[14:20:04] <inariksit> yeah :-D
[14:20:46] <inariksit> "ti" is the note
[14:20:49] <inariksit> like do re mi fa so
[14:21:06] <inariksit> comes from some weird "this must be a compound word" tokenisation
[14:21:19] <inariksit> original was finnish, and it just happened to have a word that begins with ti
[14:21:28] <iwk> Oh.
[14:21:38] <inariksit> also yarmulke was a result of an error in the finnish morphology
[14:21:49] <inariksit> "it slopes" is a genuine homonymy issue
[14:22:06] <inariksit> but then "anglicized xmas" for talking about a christmas in england
[14:22:08] <inariksit> that's just wtf
[14:22:29] <inariksit> or well, lexical selection :-P there was many entries for stuff related to england and stuff related to christmas
[14:22:30] <iwk> Yarmulke... isn't that Yiddish? O.o Or am I confusing it w/ a similar word?
[14:22:32] <inariksit> in the lexicon
[14:22:43] <inariksit> yeah, it's this jewish hat
[14:22:55] <iwk> haha, then I was right. yay me
[14:23:00] <inariksit> \o/
[14:25:20] <iwk> "en en och en en i enhet" - here GF is better, if only b/c GF has a penchant for junipers
[14:25:38] <iwk> GF: A juniper and a juniper in unit
[14:25:49] <iwk> GT: a one and one in the unit
[14:26:02] <inariksit> :--D
[14:35:36] <ehrle> "Either or to eat or one to be eate"
[14:39:58] <inariksit> nice!
[14:40:11] <inariksit> so "äta eller ätas" works fine, it is recognised as chunks
[14:40:26] <inariksit> but "antingen äta eller ätas" gets to the absolute worst, word-to-word based
[14:40:33] <ehrle> Guess the english: "Antingen vi får gift eller vi går sönder upp"
[14:40:37] <inariksit> yeah!
[14:41:20] <iwk> awesome
[14:41:28] <inariksit> get married -- får gift
[14:41:30] <inariksit> naturally :-D
[14:42:09] <iwk> it didn't hurt me, but I guess in some cases :P
[14:42:35] <inariksit> if you write "får gift" in Swe-Eng, it gives "sheep marriedly"
[14:42:43] <iwk> ROFL
[14:43:33] <iwk> but "han får gift" is actually translated as "he gets poison". Almost correct
[14:43:46] <iwk> Well, syntactically perfect, but pragmatically odd
[14:49:48] <iwk> Haha, yet another thing:
[14:49:55] <iwk> "jag hoppas att du inte tar illa upp"
[14:50:03] <iwk> Google: I hope you do not mind
[14:50:14] <iwk> GF: I hope up that you do not fork ill
[14:51:18] <iwk> (btw, that one was yellow, which I find odd)
[14:52:03] <inariksit> ehrle: inari@chalmers.se
[14:52:40] <inariksit> iwk: DictionarySwe.gf:lin fork_V = mkV "ta" "tar" "ta" "tog" "tagit" "tagen" ; -- comment=5
[14:52:56] <inariksit> that's in your batch to correct, go ahead!
[14:53:00] <inariksit> or is it correct already
[14:54:00] <iwk> Hm, looks really weird to me, but I'll have to think about obscure uses before I take it out
[14:54:23] <inariksit> yeah
[14:54:38] <inariksit> it's fine to have obscure translations for obscure verbs
[14:54:44] <inariksit> what even is an intransitive fork
[14:54:59] <inariksit> alright, now I'm off
[14:55:01] <inariksit> \o
[14:55:08] <iwk> 1. Dunno 2. Bye :)
[14:55:40] <iwk> Oh wow, is that the time
[14:55:44] <iwk> Guess I'm going too
[14:55:49] <ehrle> hej hej
[14:55:53] *** Quits: ehrle (82f13110@gateway/web/freenode/ip.130.241.49.16) (Quit: Page closed)
[14:56:12] *** Quits: iwk (~user@flv49020.hum.gu.se) (Remote host closed the connection)
[15:07:38] *** Joins: crazydiamond (~crazydiam@178.141.71.239)
[15:10:05] *** Quits: spectre (~fran@dhcp507-ans.wifi.uit.no) (Ping timeout: 244 seconds)
[15:18:37] *** Quits: lt_ (82f106c4@gateway/web/freenode/ip.130.241.6.196) (Ping timeout: 246 seconds)
[16:15:44] *** Joins: spectre (~fran@115.114.202.84.customer.cdi.no)
[17:38:53] *** Quits: vin-ivar (~vinit@122.169.38.69) (Quit: WeeChat 1.1.1)
[17:39:38] *** Joins: vin-ivar (~vinit@122.169.38.69)
[17:41:02] *** Quits: vin-ivar (~vinit@122.169.38.69) (Client Quit)
[17:41:29] *** Joins: vin-ivar (~vinit@122.169.38.69)
[17:56:47] *** Joins: BobL (02454fa9@gateway/web/freenode/ip.2.69.79.169)
[19:34:21] *** Quits: BobL (02454fa9@gateway/web/freenode/ip.2.69.79.169) (Quit: Page closed)
[20:08:10] *** Quits: jmvanel (~jmvanel@crz75-3-78-193-235-5.fbxo.proxad.net) (Remote host closed the connection)
[20:38:25] *** Quits: LukeG (82f13112@gateway/web/freenode/ip.130.241.49.18) (Quit: Page closed)