1% This file is part of the Attempto Parsing Engine (APE). 2% Copyright 2008-2013, Kaarel Kaljurand <kaljurand@gmail.com>. 3% 4% The Attempto Parsing Engine (APE) is free software: you can redistribute it and/or modify it 5% under the terms of the GNU Lesser General Public License as published by the Free Software 6% Foundation, either version 3 of the License, or (at your option) any later version. 7% 8% The Attempto Parsing Engine (APE) is distributed in the hope that it will be useful, but WITHOUT 9% ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR 10% PURPOSE. See the GNU Lesser General Public License for more details. 11% 12% You should have received a copy of the GNU Lesser General Public License along with the Attempto 13% Parsing Engine (APE). If not, see http://www.gnu.org/licenses/. 14 15 16:- module(drs_to_owldrs, [ 17 drs_to_owldrs/2 18 ]). 19 20:- use_module(ape('utils/drs_ops'), [ 21 unary_drs_operator/1 22 ]).

DRS to OWL DRS converter

The DRS is converted to "OWL DRS" which is a more suitable format for the conversion into OWL. The input must be simplified DRS (drs_to_sdrs.pl) which has been numbervard. The following steps are made:

Rewrite implication chains: A => (B => C) ~~> (A & B) => C.
Rewrite embedded implications via conjunction and negations: A => B ~~> -(A & -B)
Remove double negations: --A ~~> A
Remove toplevel be-predicate-conditions
Remove named-object-conditions, unify the corresponding discourse referents with named(ProperName)
Rewrite certain conditions and sets of conditions:

transitive (located-in) and comparative (taller than) adjectives are turned into transitive verbs,
relation/3 + be-predicate are used to construct data properties,
otherwise, relation/3 is used to construct object properties.
"which man" is turned into "what is a man"
Map attributive and predicative positive adjectives (property/3) to nouns (object/6). Supporting comparative and superlative ("richer man", "richest man") probably doesn't make sense as these require more background semantics.
@author Kaarel Kaljurand @version 2013-04-07

*/

90drs_to_owldrs(_, [], []) :- 91 !. 92 93% Double negation is removed 94% --A == A 95drs_to_owldrs(L, [-Drs | Conds], CondsAll) :- 96 drs_to_owldrs(l(L), Drs, [-DrsR]), 97 !, 98 drs_to_owldrs(L, Conds, CondsR), 99 append(DrsR, CondsR, CondsAll). 100 101drs_to_owldrs(L, [Drs | Conds], [NDrs | NConds]) :- 102 functor(Drs, Op, 1), 103 unary_drs_operator(Op), 104 !, 105 arg(1, Drs, DrsConds), 106 drs_to_owldrs(l(L), DrsConds, NDrsConds), 107 functor(NDrs, Op, 1), 108 arg(1, NDrs, NDrsConds), 109 drs_to_owldrs(L, Conds, NConds). 110 111 112% Handle sentences like: 113% For every thing X for every thing Y if X owns something that contains Y then X owns Y. 114% BUG: maybe allow this for any level, i.e. not just 0 115% BUG: this new handling changes the OWL representation of sentences 41, 115, 116, 117. 116% Also, it makes sentences 66 and 67 correctly succeed as SRWL rules. 117% Also, it would break sentence 57 "Every man likes no dog.", but we avoid this 118% by requiring the embedded then-box not to be negated. 119% But, then we reject this: For everything X for every thing Y if X likes Y then X does not hate Y. 120% (although this could be mapped to property disjointness axiom). 121% 122% BUG: Think about all this in a more systematic way. (The rewriting is definitely 123% correct, but it might harm completeness.) 124drs_to_owldrs(0, [IfBox => [EmbeddedIfBox => EmbeddedThenBox] | Conds], DrsR) :- 125 EmbeddedThenBox \= [-_], 126 !, 127 append(IfBox, EmbeddedIfBox, MergedIfAndEmbeddedIfBox), 128 drs_to_owldrs(0, [MergedIfAndEmbeddedIfBox => EmbeddedThenBox | Conds], DrsR). 129 130% 131% Toplevel implications are kept as they are. 132% 133drs_to_owldrs(0, [Drs1 => Drs2 | Conds], [Drs1R => Drs2R | CondsR]) :- 134 !, 135 drs_to_owldrs(l(0), Drs1, Drs1R), 136 drs_to_owldrs(l(0), Drs2, Drs2R), 137 drs_to_owldrs(0, Conds, CondsR). 138 139/* 140% This is handled by the double negation removal rule. 141% A => -B == -(A & B) 142drs_to_owldrs(l(L), [IfBox => [-ThenBox] | Conds], DrsR) :- 143 !, 144 append(IfBox, ThenBox, IfThenBox), 145 drs_to_owldrs(l(L), [-IfThenBox | Conds], DrsR). 146*/ 147 148 149% Embedded implications are turned into double negations. 150% A => B == -(A & -B) 151drs_to_owldrs(l(L), [IfBox => ThenBox | Conds], DrsR) :- 152 !, 153 drs_to_owldrs(l(L), [-[-ThenBox | IfBox] | Conds], DrsR). 154 155drs_to_owldrs(L, [DRS1 v DRS2 | Conds], [DRS1R v DRS2R | CondsR]) :- 156 !, 157 drs_to_owldrs(l(L), DRS1, DRS1R), 158 drs_to_owldrs(l(L), DRS2, DRS2R), 159 drs_to_owldrs(L, Conds, CondsR). 160 161% list condition that expresses 'at most', 'less than', 'exactly' 162drs_to_owldrs(L, [List | Conds], [ListR | CondsR]) :- 163 is_list(List), 164 !, 165 drs_to_owldrs(L, List, ListR), 166 drs_to_owldrs(L, Conds, CondsR). 167 168drs_to_owldrs(L, Conds, CondsR) :- 169 rewrite_relation(Conds, FilteredConds), 170 drs_to_owldrs(L, FilteredConds, CondsR). 171 172% (Fallback) 173drs_to_owldrs(L, [Cond | Conds], [Cond | CondsR]) :- 174 drs_to_owldrs(L, Conds, CondsR). 175 176 177% of-constructions with data items 178% 179% John's age is 30. 180% John's address is "Poland". 181% 182% BUG: those should be rejected I guess because data properties 183% cannot have inverses. (This can be done later, by 184% not allowing a data object to be a subject.) 185% 186% 30 is an age of John. 187% "Poland" is an address of John. 188% 189% Note that in addition to relation/3 we also remove the object/6. 190% 191rewrite_relation(Conds, [predicate(C, PropertyName, A, DataItem)-SId/'' | FilteredConds]) :- 192 remove_many([ 193 predicate(C, be, D1, DataItem)-SId/_, 194 object(D2, PropertyName, countable, na, eq, 1)-SId/_, 195 relation(D3, of, A)-SId/_ 196 ], Conds, FilteredConds), 197 D1 == D2, D2 == D3, 198 is_dataitem(DataItem), 199 !. 200 201rewrite_relation(Conds, [predicate(C, PropertyName, A, DataItem)-SId/'' | FilteredConds]) :- 202 remove_many([ 203 predicate(C, be, DataItem, D1)-SId/_, 204 object(D2, PropertyName, countable, na, eq, 1)-SId/_, 205 relation(D3, of, A)-SId/_ 206 ], Conds, FilteredConds), 207 D1 == D2, D2 == D3, 208 is_dataitem(DataItem), 209 !. 210 211 212% of-constructions without data items 213% 214% We handle differently the sentences with copula and without copula. 215% The following sentences 216% 217% John's father is Bill. 218% Bill is a father of John. 219% 220% will be represented as: 221% 222% PropertyAssertion(ObjectProperty(father), Individual(John), Individual(Bill)) 223% 224rewrite_relation(Conds, [predicate(_, RelationalNoun, A, B)-SId/'' | FilteredConds]) :- 225 remove_many([ 226 predicate(_, be, D1, B)-SId/_, 227 object(D2, RelationalNoun, countable, na, eq, 1)-SId/_, 228 relation(D3, of, A)-SId/_ 229 ], Conds, FilteredConds), 230 D1 == D2, D2 == D3, 231 !. 232 233rewrite_relation(Conds, [predicate(_, RelationalNoun, A, B)-SId/'' | FilteredConds]) :- 234 remove_many([ 235 predicate(_, be, B, D1)-SId/_, 236 object(D2, RelationalNoun, countable, na, eq, 1)-SId/_, 237 relation(D3, of, A)-SId/_ 238 ], Conds, FilteredConds), 239 D1 == D2, D2 == D3, 240 !. 241 242% But the following sentences 243% 244% John's father likes Bill. 245% Bill likes John's father. 246% 247% will be handled by keeping the object/6 in the returned list. 248% 249% PropertyAssertion(ObjectProperty(father), Individual(John), nodeID($VAR(2))) 250% ClassAssertion(Class(owl:Thing), nodeID($VAR(2))) 251% 252% Some other tests: 253% 254% - John's brother likes everybody. 255% - John's brother likes Mary. An age of the brother is 10. 256% - Everybody's age is 31. (SWRL? Currently buggy.) 257% - Everybody's address is "Poland". (SWRL? Currently buggy.) 258% - Everybody's ancestor is Adam. 259% - Everybody's ancestor is not Adam. 260% - Every part of EU is a country. 261% 262% Note that one should be able to derive from "Every part of EU is a country. Estonia is a part of EU." 263% that "Estonia is a country." without having to assert that "Estonia is a part." 264% To make it possible we currently do not create a class "part", i.e. relational nouns 265% only create properties. 266% 267% We used to return: 268% 269%== 270% object(D, RelationalNoun, countable, na, eq, 1)-Id 271%== 272% 273% But now we return: 274% 275%== 276% object(D, something, dom, na, na, na)-Id 277%== 278% 279rewrite_relation(Conds, [ 280 predicate(_, RelationalNoun, A, D1)-SId/'', 281 object(D1, something, dom, na, na, na)-SId/'' 282 | FilteredConds]) :- 283 remove_many([ 284 object(D1, RelationalNoun, countable, na, eq, 1)-SId/_, 285 relation(D2, of, A)-SId/_ 286 ], Conds, FilteredConds), 287 D1 == D2. 288 289 290% predicate/4 291% 292% Transitive and comparative adjectives (with arguments) 293% are mapped to transitive verbs. This is debatable (and not reversible). 294% BUG: We do not currently check the Degree of property/4. 295% Note that it can be one of {pos, pos_as, comp, comp_than, sup} 296% The outcome is that "John is as rich as Mary." == "John is richer than Mary." 297% which is not wanted. I.e. one should use different property names and additionally 298% assert disjointness of the properties. 299% 300% Examples: 301% 302% John is fond-of Mary. # pos 303% England is located-in UK. # pos 304% John is as rich as Mary. # pos_as 305% John is fonder-of Mary. # comp 306% John is taller than Mary. # comp_than 307% John is more rich than Mary. # comp_than 308% John is fondest-of Mary. # sup 309 310rewrite_relation(Conds, [predicate(RefPred, PropertyName, RefArg1, RefArg2)-SId/'' | FilteredConds]) :- 311 remove_many([ 312 predicate(RefPred, be, RefArg1, RefRel1)-SId/_, 313 property(RefRel2, PropertyName, _Degree, RefArg2)-SId/_ 314 ], Conds, FilteredConds), 315 RefRel1 == RefRel2. 316 317 318% Which man ... ? --> What is a man that ... ? 319% Turns plural into singular. 320% BUG: turns mass into countable (can we have 'mass' with 'which' at all?) 321% 322% BUG: does not work: Which member of Attempto is a productive-developer? 323rewrite_relation(Conds, [ 324 query(NewRef1, QueryWord)-SId/'', 325 object(WhichRef1, Name, countable, na, eq, 1)-SId/'', 326 predicate(_, be, NewRef1, WhichRef1)-SId/'' | FilteredConds]) :- 327 remove_many([ 328 query(WhichRef1, QueryWord)-SId/_, 329 object(WhichRef2, Name, _, na, _, _)-SId/_ 330 ], Conds, FilteredConds), 331 WhichRef1 == WhichRef2. 332 333 334% Attributive positive adjectives 335% E.g. Every man likes a red cat. -> Every man likes a cat that is a red. 336rewrite_relation(Conds, [ 337 object(Ref1, Noun, Count, na, Eq, Num)-SId/TId, 338 object(NewRef, Adj, countable, na, eq, 1)-SId/'', 339 predicate(_, be, Ref1, NewRef)-SId/'' 340 | FilteredConds]) :- 341 remove_many([ 342 object(Ref1, Noun, Count, na, Eq, Num)-SId/TId, 343 property(Ref2, Adj, pos)-SId/_ 344 ], Conds, FilteredConds), 345 Ref1 == Ref2. 346 347% Predicative positive adjectives 348% E.g. John is rich. -> John is a rich. 349rewrite_relation(Conds, [ 350 object(Ref1, Adj, countable, na, eq, 1)-SId/'', 351 predicate(PRef, be, Arg1, Ref1)-SId/TId 352 | FilteredConds]) :- 353 remove_many([ 354 property(Ref1, Adj, pos)-SId/_, 355 predicate(PRef, be, Arg1, Ref2)-SId/TId 356 ], Conds, FilteredConds), 357 Ref1 == Ref2.

392is_named(CondList, predicate(_, be, X, Y)-_) :- 393 member(query(A, _)-_, CondList), 394 (A == X ; A == Y), 395 !, 396 fail. 397 398% Filters out: 399% John is a man. 400% A man is a manager. 401% A man is John. 402% Note that the rule is so complex because we want to 403% avoid filtering out numbers, strings, and the like. 404/* 405is_named(_, predicate(_, be, X, Y)-_) :- 406 var(X), 407 var(Y), 408 !, 409 X = Y. 410*/ 411 412% This filters out: John is John. (probably not wanted) 413%is_named(_, predicate(_, be, named(N), named(N))-_).

`Drs`	- is Attempto DRS
`OwlDrs`	- is a modified Attempto DRS

`Level`	- denotes the level of embedding of the DRS-box
`Drs`	- is Attempto DRS
`OwlDrs`	- is a modified Attempto DRS