Skip to content
Snippets Groups Projects
Commit cfe8e1c7 authored by Albin Henriksson's avatar Albin Henriksson
Browse files

Improve tokenizer.py to have er_link_style parameter

parent ef139068
No related branches found
No related tags found
No related merge requests found
Expand all Hide whitespace changes
Inline Side-by-side
Showing
with 139220 additions and 1496 deletions
bart/pred.json
+ 1
− 1
View file @ cfe8e1c7
{"questions": [{"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q1724. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?height. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q624. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { ?uri wdt:P31 wd:Q624. }\""}}, {"query": {"sparql": " \"SELECT (COUNT( DISTINCT?uri) AS?c) WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q924. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT (COUNT( DISTINCT?uri) AS?c) WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q1724. }\""}}, {"query": {"sparql": " \"SELECT (COUNT( DISTINCT?uri) AS?c) WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { ?uri wdt:P31 wd:Q936. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { ?uri wdt:P31 wd:Q936. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { ?uri wdt:P106 wd:Q936. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT (COUNT( DISTINCT?uri) AS?date) WHERE {?date wdt:P170?date. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q539. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT (COUNT( DISTINCT?uri) AS?c) WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT (COUNT( DISTINCT?uri) AS?c) WHERE {?uri wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { ?uri wdt:P31 wd:Q936. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q936. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q924. }\""}}, {"query": {"sparql": " \"SELECT (COUNT( DISTINCT?uri) AS?c) WHERE {?uri wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q936. }\""}}, {"query": {"sparql": " \"SELECT (COUNT( DISTINCT?uri) AS?height) WHERE {?height wdt:P170?height. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q539. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q539. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q1724. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?date WHERE {?date wdt:P279* wd:Q936. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q624. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { ?uri wdt:P31 wd:Q1748. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P279* wd:Q624. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { ?uri wdt:P106 wd:Q936. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P170?uri. }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri wdt:P31/wdt:P106?uri. }\""}}]}
\ No newline at end of file
{"questions": [{"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:SaltLakeCity dbo:timeZone?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Caesar dbo:deathPlace?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?height WHERE {?height dbo:elevation?height } ORDER BY DESC(?height) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:UnitedStates_States_of_Vietnam_War } ORDER BY DESC(?uri) OFFSET 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Butch Otter dbo:governor?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Michael_Phelps dbo:award?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?date WHERE { res:Rachel_Stevens dbo:birthDate?date }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Frank Herbert dbo:profession?uri }\""}}, {"query": {"sparql": "<s><s> \"ASK WHERE { res:Taiko dbo:musicalInstrument?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:FCPorto dbo:homeStadium?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:Airline?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:European_Monarchy } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE {?uri dbo:country?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE {?uri dbo:Airport?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:San Francisco dbo:nickname?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Angela_Merkel dbo:birthName?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Berlin dbo:leader?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:currency?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:publisher?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Bill Gates dbo:country?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Jacques_Cousteau dbo:child?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Skateboard ; dbp:country res:Sweden }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:King.?uri dbp:spouse?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE {?uri a dbo:ArgentineFilm ; dbp:starring?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Michael_Jackson dbo:deathPlace?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:University }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Milepede dbo:class?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Website?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Amanda Palmer dbo:spouse?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Sea of Galilee dbo:portrayer?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?height WHERE { res:Karakoram dbo:elevation?height }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Finland dbo:EuropeanUnion?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Big Bang Theory dbo:starring?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Computer_Scientist ; dbp:award?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Harry_Potter dbo:author?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Literature.?uri dbp:award?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Lovesick dbo:starring?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Baghdad dbo:nickname?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Montenegro dbo:city?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE {?uri dbo:populationTotal?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?length WHERE {?length dbo:length?length } ORDER BY DESC(?length) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Berlin dbo:areaCode?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:IvyLeague_University dbo:almaMater?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Australia dbo:professionalSurfer?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:party?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Mars dbo:moon?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Batman dbo:name?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Elon_Musk dbo:author?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Wikileaks dbo:author?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Sean Parnell dbo:governor?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:FordModelT dbo:manufacturer?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:Carlo Giuliani dbo:deathPlace?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbr:Animal ; dbo:extinct?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:President_Lincoln dbo:spouse?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Bertrand Russell dbo:award?uri }\""}}, {"query": {"sparql": "<s> \"ASK WHERE { res:Pamela Anderson dbo:vegan?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE {?uri dbo:spacecraft?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE {?uri dbo:populationTotal?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:RobertKennedy dbo:spouse?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?number WHERE {?number dbo:numberOfStudents?number }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:IBM dbo:revenue?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:James_Bond dbo:starring?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Bart Simpson dbo:voice?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Tom_Hanks dbo:spouse?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Yenisei dbo:country?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Car ; dbp:manufacturer res:Germany }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?date WHERE { res:Michael_Jackson dbo:deathDate?date }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?height WHERE {?height dbo:elevation?height } ORDER BY DESC(?height) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Poet.?uri dbp:author?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:gangster?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Mars dbo:missionary?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Film ; Dbo:starring?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Astronaut }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Volcano.?uri dbp:country?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Tom_Cruise dbo:starring?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Dracula dbo:creator?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:English_Wikipedia dbo:creator?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:Chancellor?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Aldi dbo:owner?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Danielle_Steel dbo:author?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Socrates dbo:influenced?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Lake Chiemsee dbo:depth?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Aerospace ; dbp:company?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:Philippines dbo:professionalSurfer?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Queen Victoria dbo:spouse?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:UnitedStates dbo:castle?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Estonia dbo:spokenLanguage?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbr:Birds ; dbo:populationTotal?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE {?uri dbo:country?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Queen Elizabeth II dbo:father?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:ChemicalElement?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:leader?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:The Netherlands dbo:movie?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Lance Bass dbo:spouse?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:TUI Travel dbo:location?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Pakistan dbo:spokenLanguage?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Bill_Clinton dbo:spouse?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Apollo 11 dbo:mission res:Mission?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Indigo dbo: wavelength?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri dbo:band res:The_Ramones }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:Scarface dbo:name?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:River } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Fort Knox dbo:location?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbr:Earl ; dbo:birthPlace?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:China dbo:emperor?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Teenage Mutant Ninja Turtles dbo:name?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Piccadilly dbo:location?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:University }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Paraguay dbo:independenceDate?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?height WHERE {?height dbo:height?height } ORDER BY DESC(?height) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Abraham Lincoln dbo:deathPlace?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?date WHERE { res:Jack Wolfskin dbo:foundedDate?date }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:Air China dbo:headquarter?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Pilsner Urquell dbo:foundedYear?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?date WHERE { res:Boston_Tea_Party dbo:date?date }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbr:Animal ; dbo:critic?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:spouse?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Earth dbo:diameter?uri }\""}}, {"query": {"sparql": " \"ASK WHERE { res:Michelle_Obama dbo:spouse?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Minnesota dbo:state?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?date WHERE { res:Muhammad dbo:deathPlace?date }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Moon dbo:mosphere?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Texas dbo:governor?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:Kurosawa dbo:director?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:San Antonio dbo:battle?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:starring?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:Baguette dbo:calorie?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:Library?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Island }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:Museum }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?date WHERE { res:DeBeers dbo:foundedDate?date }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:JFK dbo:deathPlace?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Juan Carlos I dbo:spouse?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Ice_Fire dbo:author?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:The_Beatles dbo:band?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:North-Rhine Westphalia dbo:brewery?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?date WHERE { res:Princess Diana dbo:deathDate?date }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Intel dbo:foundedBy?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:PremierLeague dbo:player?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:Cat Stevens dbo:instrument?uri }\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE { res:South Carolina dbo:river?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:Rodzilla dbo:name?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:country?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:Bridge?uri }\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE { res:Facebook dbo:founder?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": "<s><s> \"SELECT DISTINCT?uri WHERE {?uri rdf:type dbo:HikingTrail.?uri dbp:country?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE {?uri dbo:numberOfPages?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:UnitedStates dbo:largestState?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}, {"query": {"sparql": " \"SELECT DISTINCT?uri WHERE {?uri dbo:Company?uri }\""}}, {"query": {"sparql": "<s> \"SELECT DISTINCT?uri WHERE { res:Columbus dbo:name?uri } ORDER BY DESC(?uri) OFFSET 0 LIMIT 1\""}}]}
\ No newline at end of file
bart/qald-10-test-predictions.csv
+ 394
− 394
View file @ cfe8e1c7
This diff is collapsed.
bart/qald-9-test-linked-tokenized.csv 0 → 100644
+ 151
− 0
View file @ cfe8e1c7
This diff is collapsed.
bart/qald-9-test-predictions.csv 0 → 100644
+ 151
− 0
View file @ cfe8e1c7
This diff is collapsed.
bart/qald-9-train-linked-tokenized.csv 0 → 100644
+ 409
− 0
View file @ cfe8e1c7
This diff is collapsed.
bart/sparql.ipynb
+ 342
− 1060
View file @ cfe8e1c7
This diff is collapsed.
data/qald-9-test-linked-tokenized.csv 0 → 100644
+ 151
− 0
View file @ cfe8e1c7
This diff is collapsed.
data/qald-9-test-linked.json
+ 32926
− 1
View file @ cfe8e1c7
This diff is collapsed.
data/qald-9-train-linked-tokenized.csv 0 → 100644
+ 409
− 0
View file @ cfe8e1c7
This diff is collapsed.
data/qald-9-train-linked.json
+ 104205
− 1
View file @ cfe8e1c7
This diff is collapsed.
data/tokenizer.py
+ 81
− 39
View file @ cfe8e1c7
# Parameters: input_file output_file
# Parameters: input_file output_file er_link_style(optional)
# Example usage: python tokenizer.py qald-9-test-linked.json qald-9-test-tokenized
# Example usage: python tokenizer.py qald-9-test-linked.json qald-9-test-tokenized REPLACE
# Note: output_file is the name of the output file without the extension
# Note: er_link_style is the style of entity linking to use.
# It can be "REPLACE", "APPEND" or blank. Default is no usage of entity links.
import sys
import json
def main():
print(sys.argv[0])
input_file = sys.argv[1] or "qald-9-test-linked.json"
output_file = sys.argv[2] or "qald-9-test-tokenized"
import json
# Decides if objects are changed to <objx> or not
TOKENIZE = False
if len(sys.argv) > 1:
input_file = sys.argv[1]
else:
print("Please provide input file")
sys.exit(1)
if len(sys.argv) > 2:
output_file = sys.argv[2]
else:
print("Please provide output file")
sys.exit(1)
if len(sys.argv) > 3:
er_link_style = sys.argv[3]
if (er_link_style != "REPLACE" and er_link_style != "APPEND"):
print("Please provide a valid entity/relationship linking style")
sys.exit(1)
else:
er_link_style = ""
prefixes = {
"http://dbpedia.org/resource/": "res:",
......@@ -36,42 +51,23 @@ def main():
out.write("text,summary\n")
for item in data["questions"]:
# Keeps track of how many objects we have replaced
obj_count = 0
question_string = item["question"][0]["string"]
query_string = item["query"]["sparql"]
entities = item.get("entities") or []
# out.write(f'\"{question_string}\", \"{query_string}\"\n')
# out.write(" ".join(["-".join(ent) for ent in entities])+"\n")
# print(query_string)
relations = item.get("relations") or []
# Find first occurence of PREFIX
prefix_index = query_string.find("PREFIX")
# While there is a PREFIX, find the next one
while prefix_index != -1:
# Find the word after PREFIX
prefixed_word = query_string[prefix_index + 7:query_string.find(" ", prefix_index + 7)]
# Find next occurence of />
end_index = query_string.find(">", prefix_index)
# Remove the prefix
query_string = query_string[:prefix_index] + query_string[end_index+2:]
if TOKENIZE:
# Find occurence of prefixed word:xyz in query and replace it with <obj{obj_count}>
obj_index = query_string.find(prefixed_word)
obj_end_index = query_string.find(" ", obj_index)
while obj_index != -1:
query_string = query_string[:obj_index] + f"<obj{obj_count}>" + query_string[obj_end_index:]
obj_index = query_string.find(prefixed_word)
obj_end_index = query_string.find(" ", obj_index)
obj_count += 1
# Find next occurence of PREFIX
prefix_index = query_string.find("PREFIX")
obj_count = 0
# Loop through all entities and repalce them with <obj{obj_count}>
# Replace all occurences of the prefix with value in prefixes
for prefix in prefixes:
......@@ -87,17 +83,63 @@ def main():
# Replace all occurences of "COUNT(" with "COUNT( "
query_string = query_string.replace("COUNT(", "COUNT( ")
if TOKENIZE:
for entity in entities:
entity_token = entity[1]
entity_index = question_string.find(entity_token)
# print(entity_token)
while entity_index != -1 and question_string[entity_index - 1] != "j":
question_string = question_string[:entity_index] + f"<obj{obj_count}>" + question_string[entity_index+len(entity_token):]
entity_index = question_string.find(entity_token)
obj_count += 1
# print("question_string", question_string)
if er_link_style == "APPEND":
# Append all entities to the end of the question
if (len(entities) > 0):
question_string += " | "
for k, entity in enumerate(entities):
if k > 0:
question_string += " | "
uri = entity["URI"]
# Hopefully the uri includes a uri which we know how to shorten
for prefix in prefixes:
if uri.startswith(prefix):
uri = prefixes[prefix] + uri[len(prefix):]
break
question_string += f"{uri}"
if er_link_style == "REPLACE":
# Helper function since entities and relations are replaced in the same way.
def replace_entity_or_relation(er_source, question_string):
for er in er_source:
uri = er["URI"].strip()
surface_form = er["surface form"].strip()
if not surface_form:
continue
# Hopefully the uri includes a uri which we know how to shorten
for prefix in prefixes:
if uri.startswith(prefix):
uri = prefixes[prefix] + uri[len(prefix):]
break
# Find the first occurence of the surface form
er_index = question_string.find(surface_form)
while er_index != -1:
# Check if the surface form is not part of an already replaced entity/relation
previous_colon = question_string.rfind(":", 0, er_index)
previous_space = question_string.rfind(" ", 0, er_index)
previous_space = max(0, previous_space)
# If there is a colon to the left and there is a space after it then it is part of an entity/relation
if previous_colon > previous_space and previous_colon != -1:
next_space = question_string.find(" ", er_index)
er_index = question_string.find(surface_form, next_space)
continue
# Else replace the surface form with the uri
question_string = question_string[:er_index] + uri + question_string[er_index+len(surface_form):]
# Find the next occurence of the surface form
er_index = question_string.find(surface_form, er_index + len(uri))
return question_string
question_string = replace_entity_or_relation(entities, question_string)
question_string = replace_entity_or_relation(relations, question_string)
out.write(f'\"{question_string}\", \"{query_string}\"\n')
if __name__ == "__main__":
......
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment