kr.angelov
2eb37f6407
bug fix in the management of memory pools in the statistical parser
2013-05-07 08:30:32 +00:00
kr.angelov
561e478ed4
the statistical parser is now using two memory pools: one for parsing and one for the output trees. This means that the memory for parsing can be released as soon as the needed abstract trees are retrieved, while the trees themselves are retained in the separate output pool
2013-05-06 15:28:04 +00:00
aarne
3e22349d2b
mkPredet in ParadigmsIta; should be more general, but this one is enough for the MGL
2013-05-06 15:25:23 +00:00
kr.angelov
307e0854ed
fix the leftcorner filtering after the addition of word completion
2013-05-05 10:30:06 +00:00
kr.angelov
be8d72d64c
bugfix in the C runtime which was causing an infinite loop while linearizing partial trees
2013-05-04 13:32:57 +00:00
normundsg
5182065168
Latvian: parameter optimisation, code clean-up and formatting, renaming
2013-05-04 04:56:46 +00:00
kr.angelov
b7ada4d269
remove /lib/src/english/ParseEngAbs3.probs since now it is moved to /treebanks/PennTreebank/ParseEngAbs3.probs
2013-05-03 21:25:31 +00:00
hallgren
5d7658c081
Download page: the .deb packages also work on Ubuntu 13.04
2013-05-03 13:14:27 +00:00
gregoire.detrez
cf5c6b18d9
[haskell runtime] Remove trailing whitespaces in VisualizeTree.hs
2013-05-03 09:42:29 +00:00
aarne
6380053fb5
started DictEngChi
2013-05-03 09:10:22 +00:00
normundsg
9f5c805ab5
Latvian: an optimized NP linearization type => fixed the double negation support in VPS
...
The flag indicating whether a NP has a negated determiner is now encapsulated in the algebraic parameter Agr, removing the extra field in the NP linearization type.
2013-05-02 13:17:53 +00:00
hallgren
0dbddbb9d7
Setup.hs: comment out ineffective parallelisation
2013-05-02 12:36:16 +00:00
normundsg
e17c87e955
Latvian extra: initial support for VP conjunction (VPS)
...
TODO: subject-dependent double negation, other verb moods (currently only the indicative mood).
2013-05-02 07:41:44 +00:00
normundsg
0660bd3779
Latvian extra: added GenCN - similarly as in ExtraFin
2013-05-01 11:08:32 +00:00
kr.angelov
9cdd96363a
word completion in the C runtime. The runtime/python/test.py example is now using readline with word completion
2013-05-01 06:09:55 +00:00
normundsg
6f4befae68
Latvian: finally handles PassV2 clauses in a more or less adequate way
...
Extends the previously introduced support for specifying the verb valence (verb-dependent subject and object cases), involving functions up to PredVP and RelVP.
In Latvian, the passive voice is not used if the agent (subject) is known; to preserve the information structure (i.e. the word order), a clause like 'A is <done> by B' is linearized in the active voice ('A <does> B') where A has the object case (e.g. Acc), and B - the subject case (e.g. Nom). Thus, the verb valence patterns are swapped on-the-fly.
This is still a rather quick & dirty implementation: parameters and linearization types have to be optimized (VerbLav), the use of PassV2 in AdvVP is problematic as it doesn't apply VPSlashPrep / ComplSlash, etc.
2013-05-01 03:33:58 +00:00
aarne
e31c92f9c3
added Mlt to make synopsis and make resourcedemo but didn't update the actual documents on the web server yet
2013-04-30 19:46:21 +00:00
john
6277bf04f9
Maltese: all functions implemented (though not necessarily correct)
...
Oh, and glad Valborg!
2013-04-30 14:23:17 +00:00
aarne
20d63050a4
missing constructs in ParseSpa, some not so certain. Present participles should be added to the morphology.
2013-04-29 13:36:48 +00:00
hallgren
d2627125e6
Download page & developers guide: add a section about known problems with Cabal
2013-04-29 13:12:37 +00:00
aarne
aac0aa2141
miniresource functor
2013-04-28 09:42:56 +00:00
kr.angelov
6cc44193b8
finally the statistical parser is able to return all possible abstract trees
2013-04-26 20:44:01 +00:00
hallgren
3d984e9ee4
gfse: use alltenses/ instead of present/ when looking up RGL oper types
2013-04-26 15:01:11 +00:00
hallgren
46546a3e93
Add greek to Setup.hs and the web apps
2013-04-26 13:42:01 +00:00
aarne
526af17c52
removed obsolete Gre files
2013-04-26 06:00:42 +00:00
aarne
563b2b185d
added Greek to ResourceDemo
2013-04-25 13:55:08 +00:00
hallgren
56f87a4c34
minibar: remove two more uses of with()
2013-04-25 15:08:26 +00:00
hallgren
debe305484
minibar: fix bug that prevented minibar from working in Webkit browsers
...
The with(...) {...} statement has different scoping rules in Webkit browsers
and Gecko browsers, so it is better to avoid it.
2013-04-25 14:55:24 +00:00
aarne
5612fcfacf
Greek by Ioanna Papodopoulou - complete, 27th RGL language
2013-04-25 13:22:28 +00:00
aarne
7190af9c7c
added Gre to lib/src/Make
2013-04-24 14:25:45 +00:00
hallgren
74ea0c9cf4
PGF service: apply the token binding operator &+ to all linearizations
...
The &+ operator is now consistently eliminated from the output of commmands
that produce linearizations. Before, only the commands translate and
translategroup did this.
2013-04-24 18:32:53 +00:00
hallgren
322968fb5c
js/localstorage.js: don't throw an exception if JSON parsing fails
...
Return the given default value instead.
2013-04-23 14:14:38 +00:00
hallgren
5c19681f2d
gfse: add access to the syntax editor
...
via the minibar
2013-04-23 13:55:00 +00:00
hallgren
dade5c1194
js/gfrobust.js: don't try to translate sentenses that are too long
...
The C run-time responds badly to input that is too long, so we return an
error message here instead.
The limit is 200 URL-encoded characters.
2013-04-23 13:40:38 +00:00
hallgren
d6c01c5414
js/localstorage.js: improved faked localStorage if real localStorage is missing
2013-04-23 12:44:12 +00:00
hallgren
aa37d9d01b
minibar: use the common interface to localStorage
...
...for more streamlined code and better browser compatibility.
2013-04-23 12:39:18 +00:00
kr.angelov
f4cf8deab7
a trivial refactoring of the reasoner in the C runtime
2013-04-23 06:40:14 +00:00
aarne
45a32c632a
new-gf-bib fixes
2013-04-22 20:48:50 +00:00
aarne
85c7571c12
Inari's new batch of words in DictFin: now 4713 checked words
2013-04-12 06:42:00 +00:00
inari.listenmaa
7868774695
new source to new-gf.bib
...
in addition to the previous ones, added also a source that isn't related to GF, but is used as a source in the sublanguage paper.
2013-04-22 17:22:11 +00:00
inari.listenmaa
697547ad40
new sources to new-gf.bib
...
Including sources needed in MTsummit paper about sublanguage. MOLTO deliverables and some conference papers.
2013-04-22 17:19:29 +00:00
normundsg
591c487d2c
Latvian: miscellaneous
2013-04-22 13:51:48 +00:00
kr.angelov
5aee2c4473
bug fix in pgf-translate which was hiding that there are more than one trees per sentence
2013-04-22 13:02:43 +00:00
kr.angelov
e76873c11f
a bit more informative error message in GrammarToPGF
2013-04-22 12:14:39 +00:00
kr.angelov
ffd64cc02a
reverse the direction of the arcs in the dependency trees
2013-04-21 19:20:08 +00:00
virk.shafqat
81bb7c7dde
refinementsDiffHindustani
2013-04-20 09:15:49 +00:00
hallgren
7c24e41d54
gfse: don't add --# -path=.:present
...
Nowadays GF includes alltenses by default in the grammar search path.
2013-04-20 07:25:38 +00:00
inari.listenmaa
038912654d
corrected error in cow_N
2013-04-19 14:47:10 +00:00
inari.listenmaa
b7b3a1160a
New Catalan feminine noun smart paradigm
...
Added a one-place constructor for words that end in "ca" or "ga" in nominative.
2013-04-19 14:32:39 +00:00
hallgren
57e06e7a6a
simple translation tool: minor documentation fix
2013-04-19 14:10:03 +00:00
