/*%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% * * $Id: convertwn20.pl,v 1.10 2007/01/10 19:40:08 mvanasse Exp $ * * Description: converts WordNet's Prolog version into RDF/OWL. * * * Author: Mark F.J. van Assem (mark@cs.vu.nl) * * Adapted and extended for SWBP WG WordNet Conversion * from original script by Maarten Menken * See http://thesauri.cs.vu.nl/ * * %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% * * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * Unpacking * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * Place in directory with subdirs 'src', 'frames', 'rdf', 'rdf/full' and * 'rdf/basic'. The src dir should * contain the WordNet Prolog source files (e.g. wn_s.pl). The 'rdf' * (sub)dirs will be used for output (converted files, e.g. 'wordnet-synset.rdf'). * The 'frames' dir should contain the specially constructed 'sen.pl' file created * from the 'frames.vrb' file in the WN/UNIX distribution. It contains clauses of * the form 'sen(SentenceNr, String)'. * * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * Software Requirements * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * * Programmed for SWI-Prolog 5.6.9 (stable release) * http://www.swi-prolog.org/ * * Programmed against WordNet Prolog Version 2.0. * http://wordnet.princeton.edu/ * * * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * To Do * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * * - better documentation * - error/debug messages outputted to files instead of standard output * - datatypes * * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * Main commands * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * * Note that it may be necessary to set Prolog's default stack sizes higher in * order for the program to run (see Settings > Stack Sizes), or run Prolog in * high mem mode. * * - go. Convert all of WordNet (not advisable for older PCs) * This creates different rdf files in the 'rdf' * directory * - go_synsets * Only does synsets, wordsenses, words * * - go_relations Does the rest * * * - create_X_file convert only the X clauses and output to file (see bottom of this file) * e.g. create_synsets_file convert only the s(...) clauses and output to file * * - create_X convert only X-clauses and assert rdf triples * e.g create_synsets convert only s(...)-clauses and assert rdf triples * - ... * * Recommended usage: * - run 'go_synsets' * - kill the current prolog process (to free up RAM) * - run 'go_relations' (frames not implemented yet) * * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * Debugging * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * * Error messages turned on by default when using the go... predicates. * Other messages are turned off. * Use 'debug(Cat)' to turn * them on. For which 'Cat' to use see the create_X predicates. E.g. for * create_pertainsto the categories 'per-types' and 'per-errors' are defined, * most follow this pattern (types: print for which sensetypes the relation * is asserted, errors: print errors such as combinations of sensetypes which * should not occur in the sourcefiles according to the documentation). * * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * CVS change log: see bottom */ % load the SWI-Prolog SemWeb libraries :- use_module(library('semweb/rdf_db')). % use_module(library('semweb/rdfs')). % Make all the WN predicates dynamic so they can be deleted after processing % for efficiency. :- dynamic s/6, g/2, hyp/2, ent/2, sim/2, mm/2, ms/2, mp/2, der/4, cls/3, cs/2, vgp/4, at/2, ant/4, sa/4, ppl/4, per/4, fr/3. %%%%%%%%%%%%%%%%%%%%% set the WN namespace %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % for testing with old code %rdf_db:ns(wn, 'http://wordnet.princeton.edu/wn20#'). %rdf_db:ns(wn20, 'http://wordnet.princeton.edu/wn20/'). % the namespaces rdf_db:ns(wn20schema, 'http://www.w3.org/2006/03/wn/wn20/schema/'). rdf_db:ns(wn20instances, 'http://www.w3.org/2006/03/wn/wn20/instances/'). % is this repeat needed? :- rdf_register_ns(wn20schema, 'http://www.w3.org/2006/03/wn/wn20/schema/'). :- rdf_register_ns(wn20instances,'http://www.w3.org/2006/03/wn/wn20/instances/'). %%%%%%%%% set the directories/filenames of input and output files %%%%%%%%%%%%%% src_dir(wn, 'src'). src_dir(fr, 'frames'). % all source files in above source dir except for the % frames file as it is created from the Unix dist, it is % not part of the src files in the normal WN Prolog dist % Dir for files that belong to both the "full" and "basic" versions out_dir(wn, 'rdf'). % Dir for files that belong to the "full" version only out_dir(wnfull, 'rdf/full'). % Dir for files that belong to the "basic" version only out_dir(wnbasic, 'rdf/basic'). % out_file/3 defines for each "kind" of output to which dir and which file it should be % outputted. % the "kind" associates source files and processing predicates out_file(sl, wnbasic, 'wordnet-senselabels.rdf'). out_file(ant, wnfull, 'wordnet-antonym.rdf'). out_file(der, wnfull, 'wordnet-derivationallyrelated.rdf'). out_file(ppl, wnfull, 'wordnet-participleof.rdf'). out_file(per, wnfull, 'wordnet-pertainsto.rdf'). out_file(sa, wnfull, 'wordnet-seealso.rdf'). out_file(ws, wnfull, 'wordnet-wordsensesandwords.rdf'). out_file(at, wn, 'wordnet-attribute.rdf'). out_file(cs, wn, 'wordnet-causes.rdf'). out_file(cls, wn, 'wordnet-classifiedby.rdf'). out_file(ent, wn, 'wordnet-entailment.rdf'). out_file(fr, wn, 'wordnet-frame.rdf'). out_file(g, wn, 'wordnet-glossary.rdf'). out_file(hyp, wn, 'wordnet-hyponym.rdf'). out_file(mm, wn, 'wordnet-membermeronym.rdf'). out_file(mp, wn, 'wordnet-partmeronym.rdf'). out_file(vgp, wn, 'wordnet-sameverbgroupas.rdf'). out_file(sim, wn, 'wordnet-similarity.rdf'). out_file(ms, wn, 'wordnet-substancemeronym.rdf'). out_file(s, wn, 'wordnet-synset.rdf'). %%%%%%%%%%%%%% construct the complete path to output file %%%%%%%%%%%%%%%%%% % F is a code used to match to actual output file, see out_file/3. out_file_path(F, OutFilePath) :- out_file(F, DirCode, File), out_dir(DirCode, Dir), concat_atom([Dir, '/', File], OutFilePath). %%%%%%%%%%%%%%%%%%%%% synset types -> synset classnames %%%%%%%%%%%%%%%%%%%%%%% synset_type(n, 'NounSynset'). synset_type(v, 'VerbSynset'). synset_type(a, 'AdjectiveSynset'). synset_type(s, 'AdjectiveSatelliteSynset'). synset_type(r, 'AdverbSynset'). %%%%%%%%%%%%%%%%%%%%% wordsense types -> wordsense classnames %%%%%%%%%%%%%%%%% % the wordsense_type is determined by the synset type (means the synset % contains only wordsenses of this type. wordsense_type(n,'NounWordSense'). wordsense_type(v,'VerbWordSense'). wordsense_type(a,'AdjectiveWordSense'). wordsense_type(s,'AdjectiveSatelliteWordSense'). wordsense_type(r,'AdverbWordSense'). %%%%%%% (word)sense types -> lexical representation used in URIs %%%%%%%%%%%%%%%%% % the URIs for Synsets and WordSenses include the lexical type, so need a mapping % to lowercase words without the addition of "Synset" or "WordSense". lexical_type_uri(n,'noun'). lexical_type_uri(v,'verb'). lexical_type_uri(a,'adjective'). lexical_type_uri(s,'adjectivesatellite'). lexical_type_uri(r,'adverb'). %%%%%%%%%%%%%%%%%%%% class_type -> classified property %%%%%%%%%%%%%%%%%%%%% class_type(t, classifiedByTopic). class_type(u, classifiedByUsage). class_type(r, classifiedByRegion). %%%%%%%%%%% Predicate used to enable dynamic forming of local:ID URIs %%%%%%%%%% % In order to put a variable after a namespace the use of rdf_global is required. my_rdf_assert(S0,P0,O0) :- rdf_global_id(S0, S), rdf_global_id(P0, P), rdf_global_term(O0, O), rdf_assert(S,P,O). %%%%%%%%%%%%%%%%%%%% Symbols in Words -> Symbols in word part in URI %%%%%%%%%%%%%%%%%%%%% uri_map([ '\'' / '_', '/' / '_', '(' / '_', ')' / '_' ]). % for each character in a word that is not allowed in a URI, replace it with % character in the uri_map % Also replaces the character if it is the last character in the word, e.g. resulting % in 'bla_bla_'. uri_word(Word, URIWord) :- % get the mapping of characters to an allowed character uri_map(L), % check for each Char in the uri map if it is present, if so replace. map_char(L, Word, URIWord). %%%%%%%%%%%%%%%%%%%% Symbols in Words -> Symbols in human-readable parts %%%%%%%%%%%%%%%%%%%%% % all we do here is replace underscores with spaces. label_map([ '_' / ' ' ]). label_word(Word, LabelWord) :- label_map(L), map_char(L, Word, LabelWord). %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Helper predicate for uri_word and label_word. % Replaces all occurences of one char (CharIn) with another (CharOut), and does % so for each provided CharIn/CharOut pair map_char([], OldWord, OldWord). map_char([CharIn/CharOut | Tail], OldWord, NewWord) :- % check if CharIn is in the word (needed because concat_atom always succeeds) sub_atom(OldWord, _, _, _, CharIn), % replace it concat_atom(List, CharIn, OldWord), concat_atom(List, CharOut, TmpWord), map_char(Tail, TmpWord, NewWord), ! % red cut, dont backtrack to deliver empty variable % if CharIn not present continue to check the next CharIn/CharOut pair ; map_char(Tail, OldWord, NewWord). %%%%%%%%%%%%%%%%%%%% collocation %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % fail if word is not a collocation in WN, i.e. does not contain '-' or '_' collocation(Word) :- sub_atom(Word,_,_,_,'-') ; sub_atom(Word,_,_,_,'_'). %%%%% Given a list of file names, load them from the src directory %%%%%%%%%%%%% load_wordnet_prolog_files(File) :- atom(File), src_dir(wn, Dir), concat_atom([Dir, File], '/', Path), consult(Path). load_wordnet_prolog_files([Head | Tail]) :- load_wordnet_prolog_files(Head), load_wordnet_prolog_files(Tail). %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % The following predicates are together responsible for converting s(_,_) % clauses to the appropriate rdf triples (Synsets, Words, WordSenses) % % - assumes wn_s.pl is loaded % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Predicates for constructing instance URIs (they produce no side effects) %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%% Example: wn20instances:word-bank %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% construct_word_uri(Word, WordInstanceURI) :- % replace disallowed characters uri_word(Word, TheWord), concat_atom(['word-', TheWord], LocalId), % use rdf_global to be able to add a variable after a namespace rdf_global_id(wn20instances:LocalId, WordInstanceURI). %%%%%%% Example: - wn20instances:wordsense-bank-noun-1 %%%%%%%%%%%%%%%%%%%%%%%%% construct_wordsense_uri(s(_Id,_Wnum, Word, SenseType, SenseNr, _TC), WordSenseInstanceURI) :- % replace disallowed characters uri_word(Word, URIWord), lexical_type_uri(SenseType, URISenseType), concat_atom(['wordsense-', URIWord, '-', URISenseType, '-', SenseNr], LocalId), % rdf_global makes possible to add variable after a namespace rdf_global_id(wn20instances:LocalId, WordSenseInstanceURI). %%%%%%%%% Example: wn20instances:synset-bank-noun-1 %%%%%%%%%%%%%%%%%%%%%%%%%%%% construct_synset_uri(s(Id,_Wnum, _Word, SenseType, _SenseNr, _TC), SynsetInstanceURI) :- % select first word(sense) of synset to use as part of the URI for the Synset instance % EDIT: removed SenseNr as constraining variable; sometimes the sensenr % does not correspond to the first word and this predicate fails... % Second edit: DO use the SenseNr that is given by the first WordSense of the synset, % just dont require it to be the SAME as the SenseNr provided in the predicate call. s(Id, 1,FirstWord,_, SenseNr,_), % replace disallowed characters uri_word(FirstWord, URIWord), lexical_type_uri(SenseType, URISenseType), concat_atom(['synset-', URIWord, '-', URISenseType, '-', SenseNr], LocalId), % rdf_global makes possible to add variable after a namespace rdf_global_id(wn20instances:LocalId, SynsetInstanceURI). %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % predicates that create triples %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % no checking of errors or doubles yet... create_synsets :- % only select the first WordSense of the synset, that is all that is required % to construct the synsets URI, synsetId and rdfs:label properties s(Id, 1,Word,SenseType,SenseNr,TagCount), % create synset instance, with synsetId and rdfs:label create_synset(s(Id, 1,Word,SenseType,SenseNr,TagCount)), % all done, move to next s(...)-es fail ; true. % do the actual asserts. % create_synset works for any s(...) provided to it. create_synset(s(Id, Wnum,Word,SenseType,SenseNr,TagCount)) :- % create URI of the right subclass of Synset (NounSynset, ...) synset_type(SenseType,SynsetClass), construct_synset_uri(s(Id,Wnum, Word, SenseType, SenseNr, TagCount), SynsetInstanceURI), my_rdf_assert(SynsetInstanceURI, rdf:type, wn20schema:SynsetClass), my_rdf_assert(SynsetInstanceURI, wn20schema:'synsetId', literal(Id)), % now add an rdfs:label by choosing the first word s(Id,1,FirstWord,_,_,_), label_word(FirstWord, LabelWord), my_rdf_assert(SynsetInstanceURI, rdfs:label, literal(LabelWord)), % debug code debug(create-synset, 'Created synset with Id ~w', [Id]). %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% create_wordsensesandwords :- s(Id, _,_,_,_,_), % Collect all s(...) with same Id (i.e. the list of wordsenses that together form a synset) bagof(s(Id,A,B,C,D,E), s(Id,A,B,C,D,E), Synset), process_synset(Synset), retractlist(Synset), % all done, move to next s(...)-es fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% process_synset([]). process_synset(s(Id,Wnum, Word, SenseType, SenseNr, TagCount)) :- % create instances of the wordt and the wordsense create_word(Word, WordInstanceURI), create_wordsense(s(Id, Wnum,Word,SenseType,SenseNr,TagCount), WordSenseInstanceURI), construct_synset_uri(s(Id, Wnum,Word,SenseType,SenseNr,TagCount), SynsetInstanceURI), % do the rdf asserts, at same time connect word to wordsense, wordsense to synset my_rdf_assert(WordSenseInstanceURI, wn20schema:word, WordInstanceURI ), my_rdf_assert(SynsetInstanceURI, wn20schema:containsWordSense, WordSenseInstanceURI), debug(ws-types, 'asserting triples for (Synset, WS, W): ~w, ~w, ~w',[SynsetInstanceURI, WordSenseInstanceURI, WordInstanceURI]). process_synset( [ s(A,B,C,D,E,F) | Tail] ) :- process_synset(s(A,B,C,D,E,F)), process_synset(Tail). % helper predicate to delete a list of facts. retractlist([]). retractlist([H|T]) :- retract(H), retractlist(T). %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% create_word(Word, WordInstanceURI) :- % A word is either of class Collocation or Word ( collocation(Word) -> WordClass = 'Collocation' ; WordClass = 'Word' ), construct_word_uri(Word, WordInstanceURI), label_word(Word, LabelWord), my_rdf_assert(WordInstanceURI, rdf:type, wn20schema:WordClass), my_rdf_assert(WordInstanceURI, wn20schema:lexicalForm, literal(LabelWord)). % create all triples for a given wordsense. Return the URI of the new instance create_wordsense(s(Id,Wnum, Word,SenseType,SenseNr,TagCount), WordSenseInstanceURI) :- construct_wordsense_uri(s(Id,Wnum, Word,SenseType,SenseNr,TagCount), WordSenseInstanceURI), wordsense_type(SenseType, ClassName), label_word(Word, LabelWord), my_rdf_assert(WordSenseInstanceURI, rdf:type, wn20schema:ClassName), my_rdf_assert(WordSenseInstanceURI, wn20schema:tagCount, literal(TagCount)), my_rdf_assert(WordSenseInstanceURI, rdfs:label, literal(LabelWord)). %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% create_senselabels :- s(Id, Wnum,Word,SenseType,SenseNr,TagCount), construct_synset_uri(s(Id, Wnum,Word,SenseType,SenseNr,TagCount), SynsetInstanceURI), label_word(Word, LabelWord), rdf_assert(SynsetInstanceURI, wn20schema:senseLabel, literal(LabelWord)), % all done, move to next s(...)-es fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % The following 'create_X' predicates each % % - assume the correct wn_Y.pl files are loaded % - convert all wordnet predicates of one type (e.g. hyp(_,_) to the % appropriate rdf triples using rdf_assert/3 % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % each synset has a natural language label (definition or gloss) create_glossaries :- g(SynsetId, Gloss), s(SynsetId,A,B,C,D,E), construct_synset_uri(s(SynsetId,A,B,C,D,E), SynsetURI), rdf_assert(SynsetURI, wn20schema:gloss, literal(Gloss)), % done with this g(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % non-symmetric relationship between synsets create_hypernyms :- hyp(Hypo, Hyper), s(Hypo,A1,B1,C1,D1,E1), s(Hyper,A2,B2,C2,D2,E2), construct_synset_uri(s(Hypo,A1,B1,C1,D1,E1), Subject), construct_synset_uri(s(Hyper,A2,B2,C2,D2,E2), Object), rdf_assert(Subject, wn20schema:hyponymOf, Object), % done with this hyp(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % non-symmetric relationship between Verb synsets create_entailments :- ent(VerbSynsetId1, VerbSynsetId2), s(VerbSynsetId1,A1,B1,C1,D1,E1), s(VerbSynsetId2,A2,B2,C2,D2,E2), construct_synset_uri(s(VerbSynsetId1,A1,B1,C1,D1,E1), Subject), construct_synset_uri(s(VerbSynsetId2,A2,B2,C2,D2,E2), Object), rdf_assert(Subject, wn20schema:entails, Object), % done with this ent(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % symmetric relation between synsets % no need to check for the inverses because it seems they are all not there % in source file (at least not for a LOT of them). create_similarities :- sim(SynsetId1, SynsetId2), s(SynsetId1,A1,B1,C1,D1,E1), s(SynsetId2,A2,B2,C2,D2,E2), construct_synset_uri(s(SynsetId1,A1,B1,C1,D1,E1), Subject), construct_synset_uri(s(SynsetId2,A2,B2,C2,D2,E2), Object), % just to be sure we check if the inverse was not already asserted, % but we wont check for 'erroneous' missing inverses \+rdf(Object, wn20schema:similarTo, Subject), rdf_assert(Subject, wn20schema:similarTo, Object), % done with this sim(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % non-symmetric relationship between synsets create_membermeronyms :- mm(SynsetId1, SynsetId2), s(SynsetId1,A1,B1,C1,D1,E1), s(SynsetId2,A2,B2,C2,D2,E2), construct_synset_uri(s(SynsetId1,A1,B1,C1,D1,E1), Subject), construct_synset_uri(s(SynsetId2,A2,B2,C2,D2,E2), Object), rdf_assert(Subject, wn20schema:memberMeronymOf, Object), fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % non-symmetric relationship between synsets create_substancemeronyms :- ms(SynsetId1, SynsetId2), s(SynsetId1,A1,B1,C1,D1,E1), s(SynsetId2,A2,B2,C2,D2,E2), construct_synset_uri(s(SynsetId1,A1,B1,C1,D1,E1), Subject), construct_synset_uri(s(SynsetId2,A2,B2,C2,D2,E2), Object), rdf_assert(Subject, wn20schema:substanceMeronymOf, Object), % done with this ms(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % non-symmetric relationship between synsets create_partmeronyms :- mp(SynsetId1, SynsetId2), s(SynsetId1,A1,B1,C1,D1,E1), s(SynsetId2,A2,B2,C2,D2,E2), construct_synset_uri(s(SynsetId1,A1,B1,C1,D1,E1), Subject), construct_synset_uri(s(SynsetId2,A2,B2,C2,D2,E2), Object), rdf_assert(Subject, wn20schema:partMeronymOf, Object), % done with this ms(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % symmetric relation between wordsenses % File contains the inverses, predicate only asserts a triple if the % inverse is present. Else it outputs error statement. create_derivationallyrelated :- der(SynsetId1, Wnum1, SynsetId2, Wnum2), s(SynsetId1,Wnum1,B1,C1,D1,E1), s(SynsetId2,Wnum2,B2,C2,D2,E2), ( %IF no inverse in the source; \+der(SynsetId2, Wnum2, SynsetId1, Wnum1) -> %THEN do nothing i.e. assume this ant(..) is erroneous % output some debug code debug(der-errors,'Error: inverse statement not present for der(~w, ~w, ~w, ~w)', [SynsetId1, Wnum1, SynsetId2, Wnum2]) %ELSE ; construct_wordsense_uri(s(SynsetId1,Wnum1,B1,C1,D1,E1), Subject), construct_wordsense_uri(s(SynsetId2,Wnum2,B2,C2,D2,E2), Object), % first check whether not already processed the inverse, then assert new triple \+rdf(Object, wn20schema:derivationallyRelated, Subject), rdf_assert(Subject, wn20schema:derivationallyRelated, Object) ), % done with this der(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % non-symmetric relationship between synsets converted into three separate % rdf properties create_classifiedby :- cls(SynsetId1, SynsetId2, Type), s(SynsetId1,A1,B1,C1,D1,E1), s(SynsetId2,A2,B2,C2,D2,E2), class_type(Type, PropertyName), rdf_db:ns(wn20schema, NS), concat_atom([NS, PropertyName], TheProperty), construct_synset_uri(s(SynsetId1,A1,B1,C1,D1,E1), Subject), construct_synset_uri(s(SynsetId2,A2,B2,C2,D2,E2), Object), rdf_assert(Subject, TheProperty, Object), % done with this cls(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % non-symmetric relationship between synsets create_causes :- cs(SynsetId1, SynsetId2), s(SynsetId1,A1,B1,C1,D1,E1), s(SynsetId2,A2,B2,C2,D2,E2), construct_synset_uri(s(SynsetId1,A1,B1,C1,D1,E1), Subject), construct_synset_uri(s(SynsetId2,A2,B2,C2,D2,E2), Object), rdf_assert(Subject, wn20schema:causes, Object), % done with this cs(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % symmetric relation between synsets % seems that the Wnums are not important as they are all 0, so we implement % this as a relation between synsets instead of wordsenses. % File contains the inverses, predicate only asserts a triple if the % inverse is present. Else it outputs error statement. create_verbgroups :- vgp(SynsetId1, _, SynsetId2, _), s(SynsetId1,A1,B1,C1,D1,E1), s(SynsetId2,A2,B2,C2,D2,E2), construct_synset_uri(s(SynsetId1,A1,B1,C1,D1,E1), Subject), construct_synset_uri(s(SynsetId2,A2,B2,C2,D2,E2), Object), % first check whether not already processed the inverse, then assert new triple \+rdf(Object, wn20schema:sameVerbGroupAs, Subject), rdf_assert(Subject, wn20schema:sameVerbGroupAs, Object), % done with this vgp(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % non-symmetric relationship between Noun and Adjective(Satellite) synsets % According to documentation the inverse relationship is also in the source % so have to check for the direction. % If there is no inverse then output error. % It is not a problem that the inverses are there because they result in the % same triples in this code. create_attributes :- at(SynsetId1, SynsetId2), s(SynsetId1,A1,B1,SenseType1,D1,E1), s(SynsetId2,A2,B2,SenseType2,D2,E2), ( %IF no inverse \+at(SynsetId2, SynsetId1) %THEN output error -> debug(at-errors, 'No inverse for at(~w, ~w) found, skipping...', [SynsetId1,SynsetId2]) ; %ELSE add the appropriate RDF triples ( % concatenate the sensetypes into PairType to be able to check % which case we are dealing with in one statement concat_atom([SenseType1,SenseType2], PairType), % Noun-Adjective PairType == 'na', construct_synset_uri(s(SynsetId1,A1,B1,SenseType1,D1,E1), Subject), construct_synset_uri(s(SynsetId2,A2,B2,SenseType2,D2,E2), Object), rdf_assert(Subject, wn20schema:attribute, Object) ; % same for Noun-AdjectiveSatellite PairType == 'ns', construct_synset_uri(s(SynsetId1,A1,B1,SenseType1,D1,E1), Subject), construct_synset_uri(s(SynsetId2,A2,B2,SenseType2,D2,E2), Object), rdf_assert(Subject, wn20schema:attribute, Object) % skip the PairTypes 'an' and 'sn', because these are inverses % that previous code already handled. ; % In every other case PairType is not one of the above, output an error message. % Check using 'member' is required because of use of "fail" later on to process all at(...) \+(member(PairType, ['na','ns','an','sn'])), debug(at-errors, 'Error, the Synsets ~w and ~w found in an at(...) do not have valid SenseTypes, respectively ~w and ~w' , [SynsetId1, SynsetId2, SenseType1, SenseType2]) ) ), % done with this at(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % symmetric relation between WordSenses. % according to documentation the symmetric relationship might also be in the % source file, i.e. contains both ant(A,B) and ant(B,A) which both result in % the same output triple. % Predicate checks if the symmetric relationship is also present % (for WN 2.0 this is not the case for some ant(...)). % If not, % it does NOT assert wn20schema:antonymOf. It then also outputs % error message using 'debug/2'. To see them assert 'debug(ant-errors)' before running. create_antonyms :- ant(SynsetId1, Wnum1, SynsetId2, Wnum2), s(SynsetId1,Wnum1,B1,C1,D1,E1), s(SynsetId2,Wnum2,B2,C2,D2,E2), ( %IF no inverse in the source; \+ant(SynsetId2, Wnum2, SynsetId1, Wnum1) -> %THEN do nothing i.e. assume this ant(..) is erroneous % output some debug code concat_atom(['Error, no symmetric ant for WordSenses (SynsetId, Wnum): ', SynsetId1, ', wnum=', Wnum1, ', ', SynsetId2, ', wnum2=', Wnum2, ' (triple not added)\n'], ErrorMsg), debug(ant-errors,ErrorMsg, []) %ELSE ; construct_wordsense_uri(s(SynsetId1,Wnum1,B1,C1,D1,E1), Subject), construct_wordsense_uri(s(SynsetId2,Wnum2,B2,C2,D2,E2), Object), % first check whether not already processed the inverse, then assert new triple \+rdf(Object, wn20schema:antonymOf, Subject), rdf_assert(Subject, wn20schema:antonymOf, Object) ), % done with this ant(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % non-symmetric relation between WordSenses, only verbs and adjectives % Not known if only FROM Verbs TO adjectives or that any combination is allowed. % When running this predicate it turns out that sa(...) is only between verbs in % this WN version (2.0). To see this assert 'debug(sa-types)' before running. create_seealso :- sa(SynsetId1, Wnum1, SynsetId2, Wnum2), s(SynsetId1,Wnum1,B1,SenseType1,D1,E1), s(SynsetId2,Wnum2,B2,SenseType2,D2,E2), construct_wordsense_uri(s(SynsetId1,Wnum1,B1,SenseType1,D1,E1), Subject), construct_wordsense_uri(s(SynsetId2,Wnum2,B2,SenseType2,D2,E2), Object), % concatenate the sensetypes into PairType to be able to check % which case we are dealing with in one statement concat_atom([SenseType1,SenseType2], PairType), ( member(PairType, ['vv','va', 'vs', 'aa','av', 'as', 'sv', 'sa']), rdf_assert(Subject, wn20schema:seeAlso, Object), % no inverse property yet... % debug code debug(sa-types, PairType, []) ; % in every other case PairType is not one of the above % (check using 'member' is required because of use of "fail" later on to process all at(...) \+(member(PairType, ['vv','va', 'vs', 'aa','av', 'as', 'sv', 'sa'])), % debug code concat_atom(['Error. Ids: ', SynsetId1, SenseType1, ' ', SynsetId2, SenseType2, '\n'], ErrorMsg), debug(sa-errors, ErrorMsg, []) ), % done with this sa(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Non-symmetric relation between WordSenses: adjective(satellite) -> verb create_participles :- ppl(SynsetId1, Wnum1, SynsetId2, Wnum2), s(SynsetId1,Wnum1,B1,SenseType1,D1,E1), s(SynsetId2,Wnum2,B2,SenseType2,D2,E2), construct_wordsense_uri(s(SynsetId1,Wnum1,B1,SenseType1,D1,E1), Subject), construct_wordsense_uri(s(SynsetId2,Wnum2,B2,SenseType2,D2,E2), Object), % concatenate the sensetypes into PairType to be able to check % which case we are dealing with in one statement concat_atom([SenseType1,SenseType2], PairType), ( member(PairType, ['av', 'sv']), rdf_assert(Subject, wn20schema:participleOf, Object), % debug code debug(ppl-types, PairType, []) ; % in every other case PairType is not one of the above % (check using 'member' is required because of use of "fail" later on to process all at(...) \+(member(PairType, ['av', 'sv'])), % debug code concat_atom(['Error. Ids: ', SynsetId1, SenseType1, ' ', SynsetId2, SenseType2, '\n'], ErrorMsg), debug(ppl-errors, ErrorMsg, []) ), % done with this ppl(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Non-symmetric relation between WordSenses, converted into two properties: % wn20schema:adjectivePertainsTo, wn20schema:adverbPertainsTo create_pertainsto :- per(SynsetId1, Wnum1, SynsetId2, Wnum2), s(SynsetId1,Wnum1,B1,SenseType1,D1,E1), s(SynsetId2,Wnum2,B2,SenseType2,D2,E2), construct_wordsense_uri(s(SynsetId1,Wnum1,B1,SenseType1,D1,E1), Subject), construct_wordsense_uri(s(SynsetId2,Wnum2,B2,SenseType2,D2,E2), Object), % concatenate the sensetypes into PairType to be able to check % which case we are dealing with in one statement concat_atom([SenseType1,SenseType2], PairType), ( member(PairType, ['an', 'aa', 'as', 'sn','sa', 'ss']), rdf_assert(Subject, wn20schema:adjectivePertainsTo, Object), % debug code debug(per-types, PairType, []) ; member(PairType, ['ra', 'rs']), rdf_assert(Subject, wn20schema:adverbPertainsTo, Object), % debug code debug(per-types, PairType, []) ; % in every other case PairType is not one of the above % (check using 'member' is required because of use of "fail" later on to process all at(...) \+(member(PairType, ['an', 'aa', 'as', 'sn','sa', 'ss','ra', 'rs'])), % debug code concat_atom(['Error. Ids: ', SynsetId1, SenseType1, ' ', SynsetId2, SenseType2, '\n'], ErrorMsg), debug(per-errors, ErrorMsg, []) ), % done with this ppl(...), do next fail ; true. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Relation between VerbWordSenses and a string. % Uses the fr(...) from the original WordNet Prolog distribution, as well as an % added sen(...) because the actual sentences are NOT in the Prolog dist. create_frames :- % a frame associates a WordSense (identified with synset Id and Wnum) % with a sentence nr fr(Id, SenNr, Wnum), % sen associates a sentence nr with a string sen(SenNr, String), s(Id,Wnum, Word, ST, SNr, TC), ( % case 1: WordSenseNr = 0, the frame sentence should be attached to all WordSenses in the Synset Wnum == 0, create_wordsense_uris(s(Id,Wnum, Word, ST, SNr, TC), WSUriList), attach_frame_to_wordsenses(WSUriList, String) ; %OR % case 2: WordSenseNr \== 0, the frame sentence should only be attached to the WordSense % identified by the Wnum Wnum \== 0, construct_wordsense_uri(s(Id,Wnum, Word, ST, SNr, TC), URI), rdf_assert(URI, wn20schema:frame, literal(String)) ), % done with this fr(...), do next fail ; true. % given a synset ID, construct a list of URIs for its WordSenses create_wordsense_uris(s(Id,_,_,_,_,_), URIList):- bagof(s(Id,A,B,C,D,E), s(Id,A,B,C,D,E), Synset), %get all s... that belong to the synset maplist(construct_wordsense_uri, Synset, URIList). % apply construct_wordsense uri to each s(...) attach_frame_to_wordsenses([], _String). attach_frame_to_wordsenses([H|T], String) :- rdf_assert(H, wn20schema:frame, literal(String)), attach_frame_to_wordsenses(T, String). %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % The following predicates are for loading the appropriate files and % subsequently calling the right create_X predicate; they form the % "main program loop" % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Create_wordnet_file: % - load the list of SrcFiles % - open the appropriate outputfile using OutFileCode % - call Predicate (which should create the RDF statements and assert them % with rdf_assert) % - save all the triples to OutFile % - add the xml:lang attribute to OutFile RDF tag with value "en-US" % - retract all triples again (for efficiency) %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% create_wordnet_file(SrcFiles, OutFileCode, Predicate) :- load_wordnet_prolog_files(SrcFiles), out_file_path(OutFileCode, OutFile), call(Predicate), debug(create-wordnet-file, 'Now going to save triples to file ~w',[OutFile]), % add lang attribute to whole doc, add namespaces to circumvent Prolog bug rdf_save(OutFile,[document_language('en-US'), namespaces([wn20instances, wn20schema])]), debug(create-wordnet-file, 'Saved triples to file, now going to retract all RDF statements',[]), rdf_retractall(_,_,_). %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Convenience predicates, each creates a separate RDF file based usually % on one WN clause % e.g. create_hypernyms_file processes all hyp(...,...). % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% create_synsets_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file('wn_s.pl', s, create_synsets). create_sense_labels_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file('wn_s.pl', sl, create_senselabels). create_wordsensesandwords_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file('wn_s.pl', ws, create_wordsensesandwords). create_glossaries_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_g.pl'], g, create_glossaries). create_hypernyms_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl', 'wn_hyp.pl'], hyp, create_hypernyms). create_entailments_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_ent.pl'], ent, create_entailments). create_similarities_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl', 'wn_sim.pl' ], sim, create_similarities). create_membermeronyms_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl', 'wn_mm.pl'], mm, create_membermeronyms). create_substancemeronyms_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_ms.pl'], ms, create_substancemeronyms). create_partmeronyms_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_mp.pl'], mp, create_partmeronyms). create_derivationallyrelated_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_der.pl'], der, create_derivationallyrelated). create_classifiedby_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_cls.pl'], cls, create_classifiedby). create_causes_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_cs.pl'], cs, create_causes). create_verbgroups_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_vgp.pl'], vgp, create_verbgroups). create_attributes_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_at.pl'], at, create_attributes). create_antonyms_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_ant.pl'], ant, create_antonyms). create_seealso_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_sa.pl'], sa, create_seealso). create_participles_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_ppl.pl'], ppl, create_participles). create_pertainsto_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), create_wordnet_file(['wn_s.pl','wn_per.pl'], per, create_pertainsto). %%%%% This predicate is different from the previous ones: %%%%% Hack needed to load the sen.pl file as it is not in the normal source %%%%% directory as is the case with all other files. %%%%% See create_frames for why this file is needed and intentionally kept separate. create_frames_file :- % turn on debug statements on file creation level so user sees % when processing is done and save to file commences. debug(create-wordnet-file), % hack to load sen.pl file from frames dir src_dir(fr, FrDir), concat_atom([FrDir, 'sen.pl'], '/', Path), consult(Path), % normal part create_wordnet_file(['wn_s.pl', 'wn_fr.pl'], fr, create_frames). %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Note that computers with limited capacity cannot handle 'go.'. % Better to: % % - run 'go_synsets' % - kill the current prolog process (to free up RAM again) % - run 'go_relations' % Especially create_wordsensesandwords_file is demanding. go :- debug_error_all, create_synsets_file, create_sense_labels_file, create_wordsensesandwords_file, create_glossaries_file, create_hypernyms_file, create_entailments_file, create_similarities_file, create_membermeronyms_file, create_substancemeronyms_file, create_partmeronyms_file, create_derivationallyrelated_file, create_classifiedby_file, create_causes_file, create_verbgroups_file, create_attributes_file, create_antonyms_file, create_seealso_file, create_participles_file, create_pertainsto_file, create_frames_file. go_synsets :- debug_error_all, create_synsets_file, create_sense_labels_file, create_wordsensesandwords_file. go_relations :- debug_error_all, create_glossaries_file, create_hypernyms_file, create_entailments_file, create_similarities_file, create_membermeronyms_file, create_substancemeronyms_file, create_partmeronyms_file, create_derivationallyrelated_file, create_classifiedby_file, create_causes_file, create_verbgroups_file, create_attributes_file, create_antonyms_file, create_seealso_file, create_participles_file, create_pertainsto_file, create_frames_file. % turn on error debug statements for all predicates that have error debug messages debug_error_all :- debug(der-errors), debug(at-errors), debug(ant-errors), debug(sa-errors), debug(ppl-errors), debug(per-errors). % fin! abode_test :- [src/wn_s], rdf_retractall(_,_,_), % Collect all s(...) with same Id (i.e. the list of wordsenses that together form a synset) %bagof(s(103141215,B,C,D,E,F), B^C^D^E^F^s(103141215,B,C,D,E,F), Synset), bagof(s(103141215,B,C,D,E,F), s(103141215,B,C,D,E,F), Synset), write(Synset), !, process_synset(Synset). /****************************************************************************** $Log: convertwn20.pl,v $ Revision 1.10 2007/01/10 19:40:08 mvanasse updated version that fixes problem with missing senses Revision 1.9 2006/05/24 13:13:35 mvanasse deleted double file name and directory for the senselabels. Reorganised the file list alphabetically Revision 1.8 2006/05/19 14:52:27 mvanasse - improved comments/documentation - fixed error in prop name: synsetContainsWordSense -> containsWordSense Revision 1.7 2006/05/17 13:12:20 mvanasse wn20schema:synsetContainsWordSense -> wn20schema:containsWordSense Revision 1.6 2006/04/25 15:22:07 mvanasse changed to new URI patterns; minor improvements in comments and structure Revision 1.5 2006/04/23 19:18:52 mvanasse *** empty log message *** Revision 1.4 2006/04/23 18:51:17 mvanasse added file for senselabels Revision 1.3 2006/04/09 18:00:12 mvanasse - removed creation of inverse properties (e.g. hypernymOf) - improved code for replacing illegal chars in URIs - improved checking for missing inverse statements and outputting error msg - more flexible routing of output to files/directories - implemented frame sentence generation - fixed bug that caused inverse wn:attribute's to be asserted Revision 1.2 2006/04/06 14:56:10 mvanasse fixed two errors in the old program: - antonym inverses are removed - language tag now en-US instead of en Revision 1.1 2006/04/06 13:33:15 mvanasse Copy of Prolog conversion program + sources belonging to the first draft (before Editor's Draft 2 February 2006). No Basic/Full version, produces the inverse properties also and old URI formatting; some bugs. - Added Princeton WN/UNIX distribution - Added dir 'frames' with frames.vrb file from the UNIX distribution - Created new 'sen.pl' file based on 'frames.vrb' *******************************************************************************/