diff options
-rw-r--r-- | sentence.cpp | 297 | ||||
-rw-r--r-- | sentence.h | 6 | ||||
m--------- | vendor/verbly | 0 |
3 files changed, 126 insertions, 177 deletions
diff --git a/sentence.cpp b/sentence.cpp index c31d226..83592d9 100644 --- a/sentence.cpp +++ b/sentence.cpp | |||
@@ -101,179 +101,23 @@ std::string sentence::generate() const | |||
101 | return verbly::implode(std::begin(words), std::end(words), " "); | 101 | return verbly::implode(std::begin(words), std::end(words), " "); |
102 | } | 102 | } |
103 | 103 | ||
104 | verbly::filter sentence::parseSelrestrs( | 104 | bool sentence::chooseSelrestr(std::set<std::string> selrestrs, std::set<std::string> choices) const |
105 | verbly::selrestr selrestr) const | ||
106 | { | 105 | { |
107 | switch (selrestr.getType()) | 106 | int validChoices = 0; |
107 | for (const std::string& choice : choices) | ||
108 | { | 108 | { |
109 | case verbly::selrestr::type::empty: | 109 | if (selrestrs.count(choice)) |
110 | { | 110 | { |
111 | return {}; | 111 | validChoices++; |
112 | } | ||
113 | |||
114 | case verbly::selrestr::type::singleton: | ||
115 | { | ||
116 | verbly::filter result; | ||
117 | |||
118 | if (selrestr.getRestriction() == "concrete") | ||
119 | { | ||
120 | result = (verbly::notion::wnid == 100001930); // physical entity | ||
121 | } else if (selrestr.getRestriction() == "time") | ||
122 | { | ||
123 | result = (verbly::notion::wnid == 100028270); // time | ||
124 | } else if (selrestr.getRestriction() == "state") | ||
125 | { | ||
126 | result = (verbly::notion::wnid == 100024720); // state | ||
127 | } else if (selrestr.getRestriction() == "abstract") | ||
128 | { | ||
129 | result = (verbly::notion::wnid == 100002137); // abstract entity | ||
130 | } else if (selrestr.getRestriction() == "scalar") | ||
131 | { | ||
132 | result = (verbly::notion::wnid == 103835412); // number | ||
133 | } else if (selrestr.getRestriction() == "currency") | ||
134 | { | ||
135 | result = (verbly::notion::wnid == 105050379); // currency | ||
136 | } else if (selrestr.getRestriction() == "location") | ||
137 | { | ||
138 | result = (verbly::notion::wnid == 100027167); // location | ||
139 | } else if (selrestr.getRestriction() == "organization") | ||
140 | { | ||
141 | result = (verbly::notion::wnid == 100237078); // organization | ||
142 | } else if (selrestr.getRestriction() == "int_control") | ||
143 | { | ||
144 | result = (verbly::notion::wnid == 100007347); // causal agent | ||
145 | } else if (selrestr.getRestriction() == "natural") | ||
146 | { | ||
147 | result = (verbly::notion::wnid == 100019128); // natural object | ||
148 | } else if (selrestr.getRestriction() == "phys_obj") | ||
149 | { | ||
150 | result = (verbly::notion::wnid == 100002684); // physical object | ||
151 | } else if (selrestr.getRestriction() == "solid") | ||
152 | { | ||
153 | result = (verbly::notion::wnid == 113860793); // solid | ||
154 | } else if (selrestr.getRestriction() == "shape") | ||
155 | { | ||
156 | result = (verbly::notion::wnid == 100027807); // shape | ||
157 | } else if (selrestr.getRestriction() == "substance") | ||
158 | { | ||
159 | result = (verbly::notion::wnid == 100019613); // substance | ||
160 | } else if (selrestr.getRestriction() == "idea") | ||
161 | { | ||
162 | result = (verbly::notion::wnid == 105803379); // idea | ||
163 | } else if (selrestr.getRestriction() == "sound") | ||
164 | { | ||
165 | result = (verbly::notion::wnid == 107111047); // sound | ||
166 | } else if (selrestr.getRestriction() == "communication") | ||
167 | { | ||
168 | result = (verbly::notion::wnid == 100033020); // communication | ||
169 | } else if (selrestr.getRestriction() == "region") | ||
170 | { | ||
171 | result = (verbly::notion::wnid == 105221895); // region | ||
172 | } else if (selrestr.getRestriction() == "place") | ||
173 | { | ||
174 | result = (verbly::notion::wnid == 100586262); // place | ||
175 | } else if (selrestr.getRestriction() == "machine") | ||
176 | { | ||
177 | result = (verbly::notion::wnid == 102958343); // machine | ||
178 | } else if (selrestr.getRestriction() == "animate") | ||
179 | { | ||
180 | result = (verbly::notion::wnid == 100004258); // animate thing | ||
181 | } else if (selrestr.getRestriction() == "plant") | ||
182 | { | ||
183 | result = (verbly::notion::wnid == 103956922); // plant | ||
184 | } else if (selrestr.getRestriction() == "comestible") | ||
185 | { | ||
186 | result = (verbly::notion::wnid == 100021265); // food | ||
187 | } else if (selrestr.getRestriction() == "artifact") | ||
188 | { | ||
189 | result = (verbly::notion::wnid == 100021939); // artifact | ||
190 | } else if (selrestr.getRestriction() == "vehicle") | ||
191 | { | ||
192 | result = (verbly::notion::wnid == 104524313); // vehicle | ||
193 | } else if (selrestr.getRestriction() == "human") | ||
194 | { | ||
195 | result = (verbly::notion::wnid == 100007846); // person | ||
196 | } else if (selrestr.getRestriction() == "animal") | ||
197 | { | ||
198 | result = (verbly::notion::wnid == 100015388); // animal | ||
199 | } else if (selrestr.getRestriction() == "body_part") | ||
200 | { | ||
201 | result = (verbly::notion::wnid == 105220461); // body part | ||
202 | } else if (selrestr.getRestriction() == "garment") | ||
203 | { | ||
204 | result = (verbly::notion::wnid == 103051540); // clothing | ||
205 | } else if (selrestr.getRestriction() == "tool") | ||
206 | { | ||
207 | result = (verbly::notion::wnid == 104451818); // tool | ||
208 | } else { | ||
209 | return {}; | ||
210 | } | ||
211 | |||
212 | std::cout << selrestr.getRestriction() << " (" << selrestr.getPos() << ")" << std::endl; | ||
213 | |||
214 | if (selrestr.getPos()) | ||
215 | { | ||
216 | return (verbly::notion::fullHypernyms %= result); | ||
217 | } else { | ||
218 | return !(verbly::notion::fullHypernyms %= result); | ||
219 | } | ||
220 | } | ||
221 | |||
222 | case verbly::selrestr::type::group: | ||
223 | { | ||
224 | std::cout << "or: " << selrestr.getOrlogic() << std::endl; | ||
225 | verbly::filter ret(selrestr.getOrlogic()); | ||
226 | |||
227 | for (const verbly::selrestr& child : selrestr) | ||
228 | { | ||
229 | ret += parseSelrestrs(child); | ||
230 | } | ||
231 | |||
232 | return ret; | ||
233 | } | ||
234 | } | ||
235 | } | ||
236 | |||
237 | bool sentence::requiresSelrestr( | ||
238 | std::string restriction, | ||
239 | verbly::selrestr selrestr) const | ||
240 | { | ||
241 | switch (selrestr.getType()) | ||
242 | { | ||
243 | case verbly::selrestr::type::empty: | ||
244 | { | ||
245 | return false; | ||
246 | } | ||
247 | |||
248 | case verbly::selrestr::type::singleton: | ||
249 | { | ||
250 | if (selrestr.getRestriction() == restriction) | ||
251 | { | ||
252 | return selrestr.getPos(); | ||
253 | } else { | ||
254 | return false; | ||
255 | } | ||
256 | } | ||
257 | |||
258 | case verbly::selrestr::type::group: | ||
259 | { | ||
260 | if (selrestr.getOrlogic()) | ||
261 | { | ||
262 | return std::all_of(std::begin(selrestr), std::end(selrestr), [=] (const verbly::selrestr& s) { | ||
263 | return requiresSelrestr(restriction, s); | ||
264 | }); | ||
265 | } else { | ||
266 | return std::any_of(std::begin(selrestr), std::end(selrestr), [=] (const verbly::selrestr& s) { | ||
267 | return requiresSelrestr(restriction, s); | ||
268 | }); | ||
269 | } | ||
270 | } | 112 | } |
271 | } | 113 | } |
114 | |||
115 | return std::bernoulli_distribution(static_cast<double>(validChoices)/static_cast<double>(selrestrs.size()))(rng_); | ||
272 | } | 116 | } |
273 | 117 | ||
274 | verbly::word sentence::generateStandardNoun( | 118 | verbly::word sentence::generateStandardNoun( |
275 | std::string role, | 119 | std::string role, |
276 | verbly::selrestr selrestrs) const | 120 | std::set<std::string> selrestrs) const |
277 | { | 121 | { |
278 | std::geometric_distribution<int> tagdist(0.5); // 0.06 | 122 | std::geometric_distribution<int> tagdist(0.5); // 0.06 |
279 | std::vector<verbly::word> result; | 123 | std::vector<verbly::word> result; |
@@ -292,11 +136,122 @@ verbly::word sentence::generateStandardNoun( | |||
292 | // Only use selection restrictions for a first attempt. | 136 | // Only use selection restrictions for a first attempt. |
293 | if (trySelection) | 137 | if (trySelection) |
294 | { | 138 | { |
295 | verbly::filter selrestrCondition = parseSelrestrs(selrestrs).compact(); | 139 | verbly::filter selection(true); |
296 | 140 | ||
297 | if (selrestrCondition.getType() != verbly::filter::type::empty) | 141 | for (const std::string& selrestr : selrestrs) |
298 | { | 142 | { |
299 | condition &= std::move(selrestrCondition); | 143 | if (selrestr == "concrete") |
144 | { | ||
145 | selection += (verbly::notion::wnid == 100001930); // physical entity | ||
146 | } else if (selrestr == "time") | ||
147 | { | ||
148 | selection += (verbly::notion::wnid == 100028270); // time | ||
149 | } else if (selrestr == "state") | ||
150 | { | ||
151 | selection += (verbly::notion::wnid == 100024720); // state | ||
152 | } else if (selrestr == "abstract") | ||
153 | { | ||
154 | selection += (verbly::notion::wnid == 100002137); // abstract entity | ||
155 | } else if (selrestr == "scalar") | ||
156 | { | ||
157 | selection += (verbly::notion::wnid == 103835412); // number | ||
158 | } else if (selrestr == "currency") | ||
159 | { | ||
160 | selection += (verbly::notion::wnid == 105050379); // currency | ||
161 | } else if (selrestr == "location") | ||
162 | { | ||
163 | selection += (verbly::notion::wnid == 100027167); // location | ||
164 | } else if (selrestr == "organization") | ||
165 | { | ||
166 | selection += (verbly::notion::wnid == 100237078); // organization | ||
167 | } else if (selrestr == "int_control") | ||
168 | { | ||
169 | selection += (verbly::notion::wnid == 100007347); // causal agent | ||
170 | } else if (selrestr == "natural") | ||
171 | { | ||
172 | selection += (verbly::notion::wnid == 100019128); // natural object | ||
173 | } else if (selrestr == "phys_obj") | ||
174 | { | ||
175 | selection += (verbly::notion::wnid == 100002684); // physical object | ||
176 | } else if (selrestr == "solid") | ||
177 | { | ||
178 | selection += (verbly::notion::wnid == 113860793); // solid | ||
179 | } else if (selrestr == "shape") | ||
180 | { | ||
181 | selection += (verbly::notion::wnid == 100027807); // shape | ||
182 | } else if (selrestr == "substance") | ||
183 | { | ||
184 | selection += (verbly::notion::wnid == 100019613); // substance | ||
185 | } else if (selrestr == "idea") | ||
186 | { | ||
187 | selection += (verbly::notion::wnid == 105803379); // idea | ||
188 | } else if (selrestr == "sound") | ||
189 | { | ||
190 | selection += (verbly::notion::wnid == 107111047); // sound | ||
191 | } else if (selrestr == "communication") | ||
192 | { | ||
193 | selection += (verbly::notion::wnid == 100033020); // communication | ||
194 | } else if (selrestr == "region") | ||
195 | { | ||
196 | selection += (verbly::notion::wnid == 105221895); // region | ||
197 | } else if (selrestr == "place") | ||
198 | { | ||
199 | selection += (verbly::notion::wnid == 100586262); // place | ||
200 | } else if (selrestr == "machine") | ||
201 | { | ||
202 | selection += (verbly::notion::wnid == 102958343); // machine | ||
203 | } else if (selrestr == "animate") | ||
204 | { | ||
205 | selection += (verbly::notion::wnid == 100004258); // animate thing | ||
206 | } else if (selrestr == "plant") | ||
207 | { | ||
208 | selection += (verbly::notion::wnid == 103956922); // plant | ||
209 | } else if (selrestr == "comestible") | ||
210 | { | ||
211 | selection += (verbly::notion::wnid == 100021265); // food | ||
212 | } else if (selrestr == "artifact") | ||
213 | { | ||
214 | selection += (verbly::notion::wnid == 100021939); // artifact | ||
215 | } else if (selrestr == "vehicle") | ||
216 | { | ||
217 | selection += (verbly::notion::wnid == 104524313); // vehicle | ||
218 | } else if (selrestr == "human") | ||
219 | { | ||
220 | selection += (verbly::notion::wnid == 100007846); // person | ||
221 | } else if (selrestr == "animal") | ||
222 | { | ||
223 | selection += (verbly::notion::wnid == 100015388); // animal | ||
224 | } else if (selrestr == "body_part") | ||
225 | { | ||
226 | selection += (verbly::notion::wnid == 105220461); // body part | ||
227 | } else if (selrestr == "garment") | ||
228 | { | ||
229 | selection += (verbly::notion::wnid == 103051540); // clothing | ||
230 | } else if (selrestr == "tool") | ||
231 | { | ||
232 | selection += (verbly::notion::wnid == 104451818); // tool | ||
233 | } else if ((selrestr == "concrete_inanimate") || (selrestr == "inanimate")) | ||
234 | { | ||
235 | selection += (verbly::notion::wnid == 100021939); // artifact | ||
236 | selection += (verbly::notion::wnid == 100019128); // natural object | ||
237 | } else if (selrestr == "non_region_location") | ||
238 | { | ||
239 | selection += (verbly::notion::wnid == 102913152); // building | ||
240 | } else if (selrestr == "non_solid_food") | ||
241 | { | ||
242 | selection += (verbly::notion::wnid == 107881800); // beverage | ||
243 | } else if (selrestr == "solid_food") | ||
244 | { | ||
245 | selection += (verbly::notion::wnid == 107555863); // solid food | ||
246 | } else if (selrestr == "slinky") | ||
247 | { | ||
248 | selection += (verbly::notion::wnid == 103670849); // line | ||
249 | } | ||
250 | } | ||
251 | |||
252 | if (selection.compact().getType() != verbly::filter::type::empty) | ||
253 | { | ||
254 | condition &= std::move(selection); | ||
300 | } else if (role == "Attribute") | 255 | } else if (role == "Attribute") |
301 | { | 256 | { |
302 | condition &= (verbly::notion::fullHypernyms %= (verbly::notion::wnid == 100024264)); // attribute | 257 | condition &= (verbly::notion::fullHypernyms %= (verbly::notion::wnid == 100024264)); // attribute |
@@ -457,7 +412,7 @@ verbly::token sentence::generateClause( | |||
457 | } | 412 | } |
458 | std::cout << std::endl; | 413 | std::cout << std::endl; |
459 | 414 | ||
460 | if (requiresSelrestr("currency", part.getNounSelrestrs())) | 415 | if (chooseSelrestr(part.getNounSelrestrs(), {"currency"})) |
461 | { | 416 | { |
462 | int lead = std::uniform_int_distribution<int>(1,9)(rng_); | 417 | int lead = std::uniform_int_distribution<int>(1,9)(rng_); |
463 | int tail = std::uniform_int_distribution<int>(0,6)(rng_); | 418 | int tail = std::uniform_int_distribution<int>(0,6)(rng_); |
@@ -559,11 +514,7 @@ verbly::token sentence::generateClause( | |||
559 | } else { | 514 | } else { |
560 | verbly::word noun = generateStandardNoun(part.getNounRole(), part.getNounSelrestrs()); | 515 | verbly::word noun = generateStandardNoun(part.getNounRole(), part.getNounSelrestrs()); |
561 | 516 | ||
562 | bool plural = part.nounHasSynrestr("plural"); | 517 | bool plural = part.nounHasSynrestr("plural") || chooseSelrestr(part.getNounSelrestrs(), {"group", "plural"}); |
563 | if (!plural) | ||
564 | { | ||
565 | plural = requiresSelrestr("plural", part.getNounSelrestrs()); | ||
566 | } | ||
567 | 518 | ||
568 | utter << generateStandardNounPhrase( | 519 | utter << generateStandardNounPhrase( |
569 | noun, | 520 | noun, |
diff --git a/sentence.h b/sentence.h index abf6860..ecdf946 100644 --- a/sentence.h +++ b/sentence.h | |||
@@ -16,11 +16,9 @@ public: | |||
16 | 16 | ||
17 | private: | 17 | private: |
18 | 18 | ||
19 | verbly::filter parseSelrestrs(verbly::selrestr selrestr) const; | 19 | bool chooseSelrestr(std::set<std::string> selrestrs, std::set<std::string> choices) const; |
20 | 20 | ||
21 | bool requiresSelrestr(std::string restriction, verbly::selrestr selrestr) const; | 21 | verbly::word generateStandardNoun(std::string role, std::set<std::string> selrestrs) const; |
22 | |||
23 | verbly::word generateStandardNoun(std::string role, verbly::selrestr selrestrs) const; | ||
24 | 22 | ||
25 | verbly::token generateStandardNounPhrase( | 23 | verbly::token generateStandardNounPhrase( |
26 | const verbly::word& noun, | 24 | const verbly::word& noun, |
diff --git a/vendor/verbly b/vendor/verbly | |||
Subproject bea3673ae1b3d19585dec56e96dbcd8a56b96e6 | Subproject e4fa0cb86d97c23c24cd7bdd62c23f03eed312d | ||