forked from opencog/relex
-
Notifications
You must be signed in to change notification settings - Fork 0
/
ChangeLog
192 lines (170 loc) · 8.69 KB
/
ChangeLog
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
New in version 1.6.3 (XXX 2016)
* Clean up wiki processing tools
* Assorted minor issues.
New in version 1.6.2 (30 Nov 2015)
* Remove Relex2Logic code; its now in OpenCog.
* Major rule updates, thanks due to Aaron Nitzkin
New in version 1.6.1 (24 Sep 2014)
* Major overhaul in comparatives, thanks to Kasun Perera.
* Major updates of Relex2Logic, multiple authors.
* Major updates for various rules.
* Bump to Java6 as minimum supported.
* Remove support for Xerces.
* Remove support for OpenNLP version 1.4 and earlier.
* Remove support for anaphora resolution (it's now in opencog).
* Remove support for lexical chunking (it doesn't belong here).
New in version 1.6.0 (14 July 2014)
* Update install-ubuntu-dependencies.sh for latest link-grammar
* Dockerfile from Danny Brian.
* Use maven to resolve dependencies, from Hendy Irawan.
New in version 1.5.1 (3 April 2014)
* Extensive fixes for comparatives, from Rodas Solomon.
* New TestRelEx.java adds more unit test cases.
* Bugfix multi-language support.
* OpenCog server can now return multiple parses.
* Rename Inflection-TAG to Subscript-TAG.
* Update Wikipedia processing scripts, including sentence splitter.
* Add new rules for extraposition processing, from Rodas.
* Add Vagrant support, for containerized build, install, run.
New in version 1.5.0 (21 November 2013)
* Minor cleanup of windows *.bat file, build and config files.
* Fix the link-grammar-server.sh file so that it works.
* Fix the opencog-server.sh file so that the port number can be set.
* Add a plain-text server, for demo'ing relex live on the web.
* Remove the obsolete OpenCogXML* output classes. (500 LOC of Java removed)
* Remove the obsolete Cerego-view classes (1.7 KLOC of Java removed)
* Remove the obsolete frame code (2.8KLOC java, 11.2 KLOC rules)
* Remove the obsolete GATE entity tagger (1.7 KLOC Java code)
* Add a new Relex2Logic output style, from Alex van der Peet.
* Allow different languages and dictionary locations to be set.
New in version 1.4.2 (24 October 2013)
* Updated maven pom file, from Krzysztof Wezowski.
* Remove obsolete parser version, which didn't build any more.
* Fix/remove assorted deprecated functions.
* Print binary relations separated from unary attributes.
* Fix "I want you to eat cake", from Rodas Solomon.
* OWL updates from Jand Hashemi (Lucky--)
New in version 1.4.1 (15 October 2012)
* Fix POS tagging for conjunctions.
* NLGen output view from Blake Lemoine
* html view from James White
* Various fixes to the Frame rules & code, from Joel Pitt
New in version 1.4.0 (21 February 2011)
* Fix conjunctions to work with current versions of link-grammar.
* Fix anaphora printing for entities.
* Fix build to look for OpenNLP-1.4 if OpenNLP-1.5 not found.
New in version 1.3.1 (15 January 2011)
* Work around JWNL bug 3006600 for infinite loop --
https://sourceforge.net/tracker/?func=detail&aid=3006600&group_id=33824&atid=409470
* Add .n-f and .n-m linkgrammar gender tag recognition
* Fix: anaphora resolution failure due to bad head-word identification.
* Fix: missing frame concept error.
* Fix: Add /usr/lib/jni to the library search path.
* Add support for OpenNLP-1.5: launchpad bug 667949
https://bugs.launchpad.net/relex/+bug/667949
New in version 1.3.0 (15 May 2010)
* New date relationships for date tagging.
* Fix: use of quantity in money expressions.
* Downcase all flag names, so as to offer a more consistent usage.
* Major restructuring of the entity maintainer interfaces
(this bumps version number)
* Fix certain morphology handlers e.g.'d->would or 'd->had
* Misc frame rules fixes from Jim White, including a crash fix
* Misc frame rules fixes from Fabricio Silva.
* Clarify JWNL not found error message problems.
New in version 1.2.1 (3 November 2009)
* Provide optional Penn-Treebank-compatible POS tagging.
* Bug fixes for misc mis-identified POS tags
* Bug fix for mis-identified advmod relation.
* Improved word features, including comparative, superlative tagging, etc.
IMPORTANT: This *requires* link-grammar-4.6.5 or newer!
* Fix crash on opencog output module, on comparative sentences.
* Fix: string substitution of "be" in frame rules (Fabricio)
* Fix: correctly pass variable nodes to opencog (Fabricio)
* Fix: printing of variables in the "simple view" (Fabricio)
* Fix: misc pet-embodiment frame rule changes (Fabricio)
* Move subscript-based feature tagging out of java code into algs file.
* Remove usage of isPastTenseForm(), in favor of verb subscripts.
* Fix: Fallback sentence splitter will use standard Java API.
New in version 1.2.0 (12 October 2009)
* Fixes DocSpliter fallback to behave when it is misconfigured.
* Fixes for anaphora resolution; head-word was missing or mis-identified.
* Fix crash when the distributed parser (remote parser) returns null.
* Don't print the phrase string in the OpenCog server, it messes the output!
* Fix issuing of _amod when superlative modifier is involved.
* Fix OpenCog Frame output to include truth values (Fabricio)
* Change OpenCog Frame output to hold frame references (Fabricio)
* Frames: add new concept vars and production rules. (Fabricio)
* Use PrepositionalRelationNode in OpenCog prep relation output.
* Tag verbs of questions as hypothetical.
* Explicitly identify "filler subjects" aka "expletives"
* Identify parataxis clauses.
* Sort parses into ranked order.
* Remove % from _%quantity relation name.
* Remove garbage generated for certain prepositional sentences.
* Remove garbage generated for certain questions, e.g. "Where is the ball?"
* Fix cases where direct, indirect object reversed.
* Fix handling of future tense in certain cases.
* Fix faulty printing of certain numeric quantities.
* Fix bug with multiple _predadj in one sentence.
* Add a prototype Stanford-parser compatibility mode.
New in version 1.1.0 (2 July 2009)
* Fix crash when generating scheme frames for opencog.
* Distinct anaphora histories for distinct conversation streams.
* Fix extents calculation in lexical chunker code.
* Requires new Java interfaces in link-grammar version 4.5.8
New in version 1.0.0 (17 June 2009)
* Add "ant help" and "ant javadoc" build targets
* Upgrade to opennlp-1.4.3, maxent-2.5.2
* Move the old, deprecated parser API to parser-old.
* Update for link-grammar version 4.5.7
* Fix _amod, _predadj, _advmod order in the frame rules.
* Fix lemma form of verb 's ("That's what the man said")
* Fix handling of word subscripts, when subscripts are multi-letter.
* Fix OpenCog output for frame elements.
* Improve POS tagging coverage.
* Strip the [?], [!] and [~] suffixes from link-grammar-guessed words.
* Add refined sentence history support for anaphora resolution.
* Export anaphoric reference output to OpenCog.
New in version 0.99.0 (May 2009)
* Fix/reverse the usage of the head-word in certain relations.
Please note that this is a MAJOR INCOMPATIBILITY from previous
versions! Please note that the frame relations are BROKEN as
a result of this change; they will be fixed later!
New in version 0.98.4 (March 2009)
* Fix build break in optional OWL-format output.
* Delete obsolete interface class RelexInfo.
* Add getPOS() method to ParsedSentence.
* Split relex-semantic-algs.txt into two files.
* Correctly identify time-related adjectives.
* Fix the way that the version number is identified during build.
* Tag more pronouns as such.
* Work around crash when input contains utf8-encoded Japanese/Chinese.
* Add opencog-server.sh for generating OpenCog output.
New in version 0.98.3 (February 2009)
* Fix bug, where the OpenNLP doc splitter was not being found.
* Clarify installation instructions.
* Tag certain date expressions as such.
* Collect sentence statistics in the CFF output parser.
* Fix parsing bug introduced in version 0.10.1
New in version 0.98.2 (January 2009)
* Fix error in optional GATE/OpenNLP build.
* Add support for automatically-propagated RelEx version number.
New in version 0.98.1: (Jaunary 2009)
* Expose the new link-grammar "disjunct string" API
(this requires link-grammar version 4.4.2).
* Add comment delimiters to the frame rules file.
* Fix, now supports jwnl-1.4rc2
* Add initial OWL-format output.
* Make install/use of GATE and OpenNLP optional.
* Cleanup README to be clearer on how to use RelEx.
New in version 0.98.0: (December 2008)
* Bump version number, in preparation for a 1.0 release.
* New JSON interface to the link-grammar parser, for improved server
performance (from Borislav Iordanov).
New in version 0.10.1:
* remove cleanSentenceString(), replace with entity id's.
* Avoid printing parens in the constituent string.
* Add a pom.xml file for use with Maven.
New in version 0.8.4:
* Convert DocSplitter to factory, so that the OpenNLP install is optional.