from rdflib import Graph, RDF, URIRef, RDFS import csv g = Graph() g.parse("lexinfo2.owl") LEXINFO = "http://www.lexinfo.net/ontology/2.0/lexinfo#" LEMON = "http://lemon-model.net/lemon#" ms_props = [] syn_args = [] relations = [] definitions = [] representations = [] usages = [] values = { "Dating": [], "Animacy": [], "Aspect": [], "Case": [], "Cliticness": [], "Definiteness": [], "Degree": [], "Finiteness": [], "Gender": [], "ModificationType": [], "Mood": [], "Negative": [], "Number": [], "AdjectivePOS": [], "AdpositionPOS": [], "AdverbPOS": [], "ConjunctionPOS": [], "DeterminerPOS": [], "ArticlePOS": [], "FusedPrepositionPOS": [], "NounPOS": [], "NumeralPOS": [], "ParticlePOS": [], "PronounPOS": [], "SymbolPOS": [], "VerbPOS": [], "PartOfSpeech": [], "Person": [], "ReferentType": [], "Tense": [], "VerbFormMood": [], "Voice": [], "TermElement": [], "TermType": [], "AbbreviatedForm": [], "Frequency": [], "NormativeAuthorization": [], "Register": [], "TemporalQualifier": [] } subjs = set(g.subjects()) def get_defn(subj): ds = list(g.objects(subj, RDFS.comment)) ds.sort() return " // ".join(ds) for subj in subjs: typed = False for typ in g.objects(subj, RDFS.subPropertyOf): if (typ == URIRef(LEXINFO + "formCaseVariant") or typ == URIRef(LEXINFO + "formDegreeVariant") or typ == URIRef(LEXINFO + "formMoodVariant") or typ == URIRef(LEXINFO + "formNumberVariant") or typ == URIRef(LEXINFO + "formPersonVariant") or typ == URIRef(LEXINFO + "formPositivityVariant") or typ == URIRef(LEXINFO + "formTenseVariant")): pass elif (typ == URIRef(LEMON + "synArg") or typ == URIRef(LEXINFO + "adjunct") or typ == URIRef(LEXINFO + "predicativeAdjunct") or typ == URIRef(LEXINFO + "clausalArg") or typ == URIRef(LEXINFO + "complement") or typ == URIRef(LEXINFO + "copulativeArg") or typ == URIRef(LEXINFO + "object") or typ == URIRef(LEXINFO + "adpositionalObject") or typ == URIRef(LEXINFO + "subject")): kind = str(typ)[len(LEXINFO):] if str(subj).endswith("copulativeSubject"): kind = "copulativeArg,subject" syn_args.append({ "id": subj.n3()[(len(LEXINFO)+1):-1], "defn": get_defn(subj), "range": str(next(g.objects(subj, RDFS.range), LEXINFO))[(len(LEXINFO)):], "kind": kind }) typed = True elif (typ == URIRef(LEMON + "lexicalVariant") or typ == URIRef(LEXINFO + "derivedForm") or typ == URIRef(LEXINFO + "contractionFor")): if typ == URIRef(LEMON + "lexicalVariant"): kind = "" else: kind = str(typ)[len(LEXINFO):] relations.append({ "id": subj.n3()[(len(LEXINFO)+1):-1], "defn": get_defn(subj), "domain": "LexicalEntry", "range": "LexicalEntry", "kind": kind }) typed = True elif (typ == URIRef(LEMON + "broader") or typ == URIRef(LEMON + "equivalent") or typ == URIRef(LEMON + "incompatible") or typ == URIRef(LEMON + "narrower") or typ == URIRef(LEMON + "senseRelation") or typ == URIRef(LEXINFO + "partitiveRelation") or typ == URIRef(LEXINFO + "holonymTerm") or typ == URIRef(LEXINFO + "meronymTerm") or typ == URIRef(LEXINFO + "relatedTerm")): if typ == URIRef(LEMON + "senseRelation"): kind = "" else: kind = str(typ)[len(LEXINFO):] relations.append({ "id": subj.n3()[(len(LEXINFO)+1):-1], "defn": get_defn(subj), "domain": "LexicalSense", "range": "LexicalSense", "kind": kind }) typed = True elif (typ == URIRef(LEMON + "definition")): definitions.append({ "id": subj.n3()[(len(LEXINFO)+1):-1], "defn": get_defn(subj), }) typed = True elif (typ == URIRef(LEXINFO + "morphosyntacticProperty") or typ == URIRef(LEMON + "property")): ms_props.append({ "id": subj.n3()[(len(LEXINFO)+1):-1], "defn": get_defn(subj), }) typed = True elif typ == URIRef(LEMON + "representation"): representations.append({ "id": subj.n3()[(len(LEXINFO)+1):-1], "defn": get_defn(subj), }) typed = True elif typ == URIRef(LEMON + "context"): usages.append({ "id": subj.n3()[(len(LEXINFO)+1):-1], "defn": get_defn(subj), }) typed = True for typ in g.objects(subj, RDF.type): if typ == URIRef('http://www.w3.org/2002/07/owl#Class'): typed = True elif (str(typ).startswith(LEXINFO) and str(typ)[len(LEXINFO):] in values): if str(typ).endswith("POS"): kind = str(typ)[len(LEXINFO):-3] if str(subj).endswith("Article") or str(subj) == LEXINFO + "article": kind = "Article" val_id = subj.n3()[(len(LEXINFO)+1):-1] if [x for x in values["PartOfSpeech"] if x["id"] == val_id]: [x for x in values["PartOfSpeech"] if x["id"] == val_id][0]["kind"] = kind else: values["PartOfSpeech"].append({ "id": val_id, "defn": get_defn(subj), "kind": kind }) elif str(typ).endswith("AbbreviatedForm"): kind = str(typ)[len(LEXINFO):] val_id = subj.n3()[(len(LEXINFO)+1):-1] if [x for x in values["TermType"] if x["id"] == val_id]: [x for x in values["TermType"] if x["id"] == val_id][0]["kind"] = kind else: values["TermType"].append({ "id": val_id, "defn": get_defn(subj), "kind": kind }) else: val_id = subj.n3()[(len(LEXINFO)+1):-1] if not [x for x in values[str(typ)[len(LEXINFO):]] if x["id"] == val_id]: values[str(typ)[len(LEXINFO):]].append({ "id": subj.n3()[(len(LEXINFO)+1):-1], "defn": get_defn(subj), }) typed = True if (not typed and subj.n3().startswith("<" + LEXINFO) and not subj.n3().endswith("Form>") and not subj.n3().endswith("Variant>")): print("Could not understand " + subj.n3()) with open("data/morphosyntactic_properties.csv", "w") as out: writer = csv.writer(out) writer.writerow(["ID","Definition"]) ms_props.sort(key=lambda x: x["id"]) for ms_prop in ms_props: if ms_prop["id"] != "morphosyntacticProperty": writer.writerow([ms_prop["id"],ms_prop["defn"]]) with open("data/syntactic_arguments.csv", "w") as out: writer = csv.writer(out) writer.writerow(["ID","Subproperty Of","Definition"]) syn_args.sort(key=lambda x: x["id"]) for syn_arg in syn_args: writer.writerow([syn_arg["id"],syn_arg["kind"],syn_arg["defn"]]) with open("data/relations.csv", "w") as out: writer = csv.writer(out) writer.writerow(["ID","Subproperty Of","Target","Definition"]) relations.sort(key=lambda x: x["id"]) for rel in relations: writer.writerow([rel["id"],rel["kind"],rel["domain"],rel["defn"]]) with open("data/definitions.csv", "w") as out: writer = csv.writer(out) writer.writerow(["ID","Definition"]) definitions.sort(key=lambda x: x["id"]) for defn in definitions: writer.writerow([defn["id"],defn["defn"]]) with open("data/representations.csv", "w") as out: writer = csv.writer(out) writer.writerow(["ID","Definition"]) representations.sort(key=lambda x: x["id"]) for rep in representations: writer.writerow([rep["id"],rep["defn"]]) with open("data/usages.csv", "w") as out: writer = csv.writer(out) writer.writerow(["ID","Definition"]) usages.sort(key=lambda x: x["id"]) for usage in usages: writer.writerow([usage["id"],usage["defn"]]) for k,v in values.items(): if len(v) > 0: with open("data/values/%s.csv" % k, "w") as out: writer = csv.writer(out) writer.writerow(["ID","Type","Definition"]) v.sort(key=lambda x: x["id"]) for w in v: writer.writerow([w["id"],w.get("kind",""),w["defn"]])