-
Notifications
You must be signed in to change notification settings - Fork 3
/
RELEASE_NOTES
192 lines (154 loc) · 7.32 KB
/
RELEASE_NOTES
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
===== Dialogs module Release Notes =====
This file contains a detailled summary of Dialogs module evolution.
For more details, please refer to the GIT log.
Dialogs 0.8 - 18/12/2012
========================
Selected features:
----------------------
- recognize forward as an adverb
- correct semantics of indirect objects in actions: when no thematic role is explicitely found, we say the action 'involves' a new entity which is qualified by the indirect cmplt.
- improved management of location adverbs: we are now able to deal with sentences like 'Look down', 'Move up'...
- separate 'pick' and 'get' in thematic roles
- associate an 'InterrogativeState' to question; delay answer to human (this commit changes the way questions are answered)
- re-enable storing of emotions in the knowledge base
- added 'stop' and 'test' as recognized action verbs
Other Changes:
--------------
- improved setup.py, updated README + minor fixes
Dialogs 0.7 - 18/07/2012
========================
Selected new features:
----------------------
- when resolving a concept with an undefinite quantifier, first try to
selected an object visible to the speaker
- to improve usability 'this' is resolved with 'pointsAt' instead
of 'focusesOn'
- Use of discrimination routines to generate relevant descriptions
of objects when answering a question
- Verbalisation of agreements after recognition of desires
- Re-enable processing of statements like 'an apple is a fruit'
(note that adjectives are discarded, like 'green' in 'a green
apple is a fruit')
- Improved discrimination output (no changes to the algorithm)
- Several improvements to what Dialogs says to the user (like 'I have
no answer' repalced by 'I don't know')
- better behaviour when Dialogs is run without a knowledge base
- Allow a list of 'standard' compound nouns in share/nouns that are
identified as such at pre-processing stage.
Experimental features:
----------------------
- preliminary support for emotion expression: depending on the success of the
grounding, Dialogs assert some emotional state in the knowledge base.
Note that this feature is disabled by commit ee3818ec because it causes some
(strange) issues to the Pellet reasoner.
Language features:
------------------
- Added support for the 'another' construct
- Added thematic roles for 'throw', 'bring'
- Added synonyms for 'go' and 'give'
- Minor other improvements (more adjectives are recognized, etc)
Other changes:
--------------
- Fixed a silly bug in pronoun reversal (you -> me)
- Labels are not created anymore when processing questions
- Many bugfixes
- Many refactoring, style & comment improvements. Much remain to be done :-)
Dialogs 0.6 - 19/10/2011
========================
Major new features:
-------------------
- concepts that are currently manipulated are marked as 'ActiveConcept' in
the ontology. This allows for instance dynamic vizualisation of the
grounding process.
- added infrastructure to assess the grammatical validity of a sentence
- added mechanism to render sentences with different formatter
(included coloured tty + CSS ones)
- termination sentences (like 'goodbye') are recognized and Dialogs exits.
- added optional support of speech recognition via gSpeet (python bindings
for the Google Speech recognition API)
- Dialogs now uses the new KB API of oro-server ('revise' action + policies)
Language features:
------------------
- Support for 'ALL' quantification
- Better support for 'SOME' quantification (when several candidates are
available, a random one is picked)
- Fixed handling of "this" + category (the category was previously ignored)
- Added support for "in front of" construct
- Added support for "everything" ('everything' is currently equivalent to
'all artifacts')/"anything"
- Some irregular nouns and adjectives have been added
- Some thematic roles have been added ('clean', 'look', 'uncover')
Grounding features:
-------------------
- Added a visibility check in the discrimination routine
- When an agent expresses a state, use an indirection with 'experience'.
For instance: I'm cold -> [myself experience GD3s3, GD3s3 hasFeature cold]
Other new features:
-------------------
- host/port of the ontology server is not hardcoded anymore (!)
- fixed UTF8 issues
- continuous refactoring of the code
- preprocessor and parser are now splitted
- fixed some logging issues
- added a failfast options for the unittests + a diff with expected results
- added a 'demo' generator that output parsing unittest to HTML
- lots of small bug fixes
- completed the documentation with more examples from the user-study and
main algorithms
Dialogs 0.5 - 09/02/2011
========================
- Renaming from dialog to dialogs
Dialogs 0.4 - 25/01/2011
========================
- added special command 'forget it' to leave an interaction
- handle 'to know':
- for instance, 'do you know the red bottle?'
- 'what object do you know?'
- 'i know the yellow banana'
- added a 'demo' mode that wait between each step (except it doesn't
because I couldn't find a way to actually wait for a keypress... cf commit
4cba0223)
- better processing of directions
- parsing support for more temporal preposistions (as soon as, until...)
- added 'X is Y' -> 'X sameAs Y'
- added the ability to 'teach' new concepts (commit f693602e):
-> 'Learn that the bottle is on the table'
- more work on the merge function (when the human complete the informations)
- parsing (but no futher handling) of time data (months, days...)
- added an explicit 'SentenceFactory' to easily build common sentences like
questions
- added verb 'to think to' in the list of known verbs
and as usual, tenth of bug fixes :-)
Dialogs 0.3 - 08/09/2010
========================
The "Moving to London" scenario pass!
Dialogs 0.2 - 12/08/2010
========================
Main features:
--------------
- input/output module to chat with the robot
- grammar-free parsing module for English natural language (NL)
- verbalizing module to reconstruct a NL sentence
- glue-ing chat module with parser + verbalizer
- resolution of personal references + anaphors
- resolution noun phrases (= identification of concepts)
- resolution action verbs based on thematic roles (dictionnary-based)
- content analysis (done for imperative sentences, statements, questions)
- statement builder (convert a parsed sentence into RDF statements)
- question handler (for question, try to answer them with ORO)
- glue-ing parsing + resolution + analysis + handling
Other features:
---------------
- handling of groups with only an adjective
- "this bottle is 'next to' the table" -> pre-processing to replace "next to" by "next+to"
- Running dialog even if no ontology server.
- Generate "subclassOf" statements:
E.g: Apples are Fruits
E.g: an Apple is a fruit
- Manage Verb Tense in a naive way ("I went to Toulouse" -> [event1205 eventOccurs PAST]).
- Process adverbs ("I go slowly to Toulouse" -> [event1205 actionSupervisionMode SLOW])
- Process quantifiers: some, all, digits (1,2,3...), none
- Handling of conjunctions (AND, OR)
- pre content-analysis -> "what color is the tape?" -> don't use the color to discriminate
- use 'safeAdd' method to detect the inconsistencies
- Coloured output for easy explaination of the Dialog module behaviour