[01:25:39] *** Quits: jmvanel (~jmvanel@76.57.127.78.rev.sfr.net) (Ping timeout: 245 seconds) [07:51:16] *** Joins: jmvanel (~jmvanel@76.57.127.78.rev.sfr.net) [08:46:03] *** Quits: jmvanel (~jmvanel@76.57.127.78.rev.sfr.net) (Ping timeout: 256 seconds) [11:38:03] *** Joins: atoll (~atoll@crossing1.epfl.ch) [11:38:07] Hi [11:38:16] how should i handle homonyms ? [11:39:38] I want them to be parser on the same fun [11:39:42] parsed* [12:01:24] also I have a problem overloading an oper [12:04:43] http://paste.isomorphis.me/k45 [13:44:07] *** Quits: atoll (~atoll@crossing1.epfl.ch) (Ping timeout: 256 seconds) [14:44:09] *** Joins: atoll (~atoll@crossing1.epfl.ch) [15:55:01] hi atoll — I don't understand your problem with homonyms. Normally you would want them to parsed as potentially multiple funs, i.e. preserving the ambiguity that exists in NL [15:55:22] oh sweet someone :D [15:55:39] for your other problem, try this: [15:55:44] mkC : Component = overload { [15:55:53] I have resolvent my other problem following the doc [15:56:42] I use Compo in my oper which is defined as Compo : Type = { s : Str } [15:56:47] (yeah sorry about the delay – I am always logged in to the channel but it might be a while until I see some posts) [15:56:56] then in lincat i define Component as Compo [15:57:31] ah right [15:57:46] great! [15:57:57] Also I have less technical questions, but my use is not translating but understanding languages. Should I try to make the informations goes up ? [15:58:03] (in the parsing) [15:58:20] like by exemple, in "the database Oracle" [15:58:33] I use Database -> Name -> Database [15:58:47] Should I add the Oracle information in the result database ? [15:58:51] should i forget it ? [15:59:04] or should i write a composite type NamedDatabase ? [16:00:17] Because I saw different "styles" in the documentation and i'm a bit lost [16:00:51] which example are you referring to? [16:04:21] It seems i had understood wrong so that's fine. But could you enlighten me on how you would represent such things ? [16:04:25] It's for camel usage [16:04:47] I have Route and Component called Entreprise Integration Pattern [16:05:12] Among these component i have things called Endpoint [16:05:27] Among these endpoint i have database [16:05:48] A database can be called "the database", or "the database Oracle" [16:06:23] (oracle can be replace with any other type of database) [16:06:55] would you define database as cat or only as fun ? [16:09:45] it seems to me that Endpoint is a cat, and database is a function which returns an Endpoint [16:09:47] e.g. [16:10:01] cat Endpoint ; [16:10:01] fun database : Str -> Endpoint ; [16:10:29] or you could have [16:10:48] fun database_oracle, databse_mysql, ... : Endpoint ; [16:11:08] if you don't want to allow open strings [16:11:57] I see [16:13:30] in the Endpoint should i define endpoint in the concrete syntax as str and put "the database oracle" in it, or only "database" ? I lack the "good practice" :D [16:15:36] you can use an oper, something liek this [16:16:10] oper mkDatabase : Str -> Endpoint = \name -> {s="the database" ++ name} [16:16:30] and then: [16:16:30] fun database_oracle = mkDatabase "Oracle" [16:17:43] I see [16:18:01] Thanx you very much [16:18:11] I don't understand why gf is not more popular [16:18:45] one day maybe :) [16:18:49] The other frameework which use more or less statistical NLP are just so hard to use in real life :D [16:19:33] When you're IBM it's probably possible, when you have a 2 month internship it seems harder [16:19:35] :D [16:22:10] we like the functional approach! but just like Haskell, functional programming still isn't really "mainstream" [16:24:14] also I had another question. How would you handle orthographic correction ? [16:25:28] Because gf is a bit rigid, thus it seems you make a simple mistake and he is totally lost. I guess you could fix that with a simple manhattan distance algorithm, but i'm sure the information you get in the parsing would make that correction more precise. [16:27:08] I guess I should use the "minibar" exemple to apply the correction on the list given :D [16:28:45] to my knowledge, no one has really worked on that much [16:29:26] although as you say, since you are provided with the possible correct completion, it should be easy to have a layer on top of that which tries to correct input errors [16:29:32] based on the valid completions [16:30:45] I guess that could be my contribution to this beautiful projet [16:32:03] Also, I have corrected some compatibility mistake in the JPGF project to be able to compile it with the latest sbt and scala-library [16:32:07] if ever you're interessted [16:32:26] (compatibility with scala 2.10) [16:32:42] you should issue a pull request on the Github project! [17:30:37] https://github.com/Crossing-Tech/JPGF Here is a small fork. I have not rewrote the test but it export jars ! [17:34:18] jstar: Feel freely to share this, it's very small modification but i guess the usage of latest scala speed it up a bit. My enterprise refuse that I dedicate too much time to contribute but i will probably adapt the tests soon [17:59:18] *** Quits: atoll (~atoll@crossing1.epfl.ch) (Ping timeout: 265 seconds) [18:10:54] *** Joins: atoll (~atoll@crossing1.epfl.ch) [18:16:00] *** Joins: jmvanel (~jmvanel@AAubervilliers-652-1-122-215.w90-3.abo.wanadoo.fr) [18:18:53] *** Quits: atoll (~atoll@crossing1.epfl.ch) (Ping timeout: 245 seconds) [18:22:48] *** Joins: atoll (~atoll@crossing1.epfl.ch) [18:23:01] GrammarToPGF.mkFId: missing category Str :< [18:28:16] *** Joins: atoll1 (~atoll@crossing1.epfl.ch) [18:28:19] jstar: any idea ? [18:28:29] *** Quits: atoll (~atoll@crossing1.epfl.ch) (Read error: Connection reset by peer) [18:33:32] *** Quits: atoll1 (~atoll@crossing1.epfl.ch) (Ping timeout: 248 seconds) [18:35:28] *** Joins: atoll1 (~atoll@crossing1.epfl.ch) [18:38:28] it seems i can't use Str as a parameter to a fun ? [18:47:58] It had to use String and not Str [18:49:27] I think you can if you use the 'literal' flag [19:06:04] *** Quits: jmvanel (~jmvanel@AAubervilliers-652-1-122-215.w90-3.abo.wanadoo.fr) (Ping timeout: 248 seconds) [19:11:51] *** Joins: jmvanel (~jmvanel@AAubervilliers-652-1-122-215.w90-3.abo.wanadoo.fr) [19:24:22] *** Quits: jmvanel (~jmvanel@AAubervilliers-652-1-122-215.w90-3.abo.wanadoo.fr) (Ping timeout: 264 seconds) [19:29:43] *** Quits: atoll1 (~atoll@crossing1.epfl.ch) (Ping timeout: 245 seconds) [21:54:50] *** Joins: jmvanel (~jmvanel@76.57.127.78.rev.sfr.net)