31 lines
23 KiB
XML
31 lines
23 KiB
XML
<configuration>
|
||
<property>
|
||
<name>jobTracker</name>
|
||
<value>yarnRM</value>
|
||
</property>
|
||
<property>
|
||
<name>nameNode</name>
|
||
<value>hdfs://nameservice1</value>
|
||
</property>
|
||
<property>
|
||
<name>oozie.use.system.libpath</name>
|
||
<value>true</value>
|
||
</property>
|
||
<property>
|
||
<name>oozie.action.sharelib.for.spark</name>
|
||
<value>spark2</value>
|
||
</property>
|
||
<property>
|
||
<name>hive_metastore_uris</name>
|
||
<value>thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083</value>
|
||
</property>
|
||
<property>
|
||
<name>hive_db_name</name>
|
||
<value>openaire</value>
|
||
</property>
|
||
<property>
|
||
<name>dedupConf</name>
|
||
<value>{"wf":{"threshold":"0.9","dedupRun":"001","entityType":"organization","orderField":"legalname","queueMaxSize":"2000","groupMaxSize":"50","slidingWindowSize":"200","idPath":".id","rootBuilder":["organization","projectOrganization_participation_isParticipant","datasourceOrganization_provision_isProvidedBy"],"includeChildren":"true"},"pace":{"clustering":[{"name":"sortedngrampairs","fields":["legalname"],"params":{"max":2,"ngramLen":"3"}},{"name":"suffixprefix","fields":["legalname"],"params":{"max":1,"len":"3"}},{"name":"urlclustering","fields":["websiteurl"],"params":{}},{"name":"keywordsclustering","fields":["legalname"],"params":{"max":2,"windowSize":4}}],"strictConditions":[{"name":"exactMatch","fields":["gridid"]}],"conditions":[{"name":"DomainExactMatch","fields":["websiteurl"]},{"name":"exactMatch","fields":["country"]}],"model":[{"name":"country","algo":"Null","type":"String","weight":"0","ignoreMissing":"false","path":".country.classid"},{"name":"legalshortname","algo":"JaroWinklerNormalizedName","type":"String","weight":"0.1","ignoreMissing":"true","path":".legalshortname.value"},{"name":"legalname","algo":"JaroWinklerNormalizedName","type":"String","weight":"0.9","ignoreMissing":"false","path":".legalname.value","params":{"windowSize":4,"threshold":0.7}},{"name":"websiteurl","algo":"Null","type":"URL","weight":"0","ignoreMissing":"true","path":".websiteurl.value","params":{"host":0.5,"path":0.5}},{"name":"gridid","algo":"Null","type":"String","weight":"0.0","ignoreMissing":"true","path":".pid[] | select(.qualifier.classid==\"grid\") | .value"}],"blacklists":{"legalname":[]},"synonyms":{"key::1":["university","università","università studi","universitario","universitaria","université","universitaire","universitaires","universidad","universitade","Universität","universitaet","Uniwersytet","университет","universiteit","πανεπιστήμιο","universitesi","universiteti"],"key::2":["studies","studi","études","estudios","estudos","Studien","studia","исследования","studies","σπουδές"],"key::3":["advanced","superiore","supérieur","supérieure","supérieurs","supérieures","avancado","avancados","fortgeschrittene","fortgeschritten","zaawansowany","передовой","gevorderd","gevorderde","προχωρημένος","προχωρημένη","προχωρημένο","προχωρημένες","προχωρημένα","wyzsza"],"key::4":["institute","istituto","institut","instituto","instituto","Institut","instytut","институт","instituut","ινστιτούτο"],"key::5":["hospital","ospedale","hôpital","hospital","hospital","Krankenhaus","szpital","больница","ziekenhuis","νοσοκομείο"],"key::6":["research","ricerca","recherche","investigacion","pesquisa","Forschung","badania","исследования","onderzoek","έρευνα","erevna","erevnas"],"key::7":["college","collegio","université","colegio","faculdade","Hochschule","Szkoła Wyższa","Высшая школа","universiteit","κολλέγιο"],"key::8":["foundation","fondazione","fondation","fundación","fundação","Stiftung","Fundacja","фонд","stichting","ίδρυμα","idryma"],"key::9":["center","centro","centre","centro","centro","zentrum","centrum","центр","centrum","κέντρο"],"key::10":["national","nazionale","national","nationale","nationaux","nationales","nacional","nacional","national","krajowy","национальный","nationaal","nationale","εθνικό"],"key::11":["association","associazione","association","asociación","associação","Verein","verband","stowarzyszenie","ассоциация","associatie"],"key::12":["society","societa","société","sociedad","sociedade","gesellschaft","społeczeństwo","общество","maatschappij","κοινωνία"],"key::13":["international","internazionale","international","internacional","internacional","international","międzynarodowy","Международный","internationaal","internationale","διεθνής","διεθνή","διεθνές"],"key::14":["community","comunita","communauté","comunidad","comunidade","Gemeinschaft","społeczność","сообщество","gemeenschap","κοινότητα"],"key::15":["school","scuola","école","escuela","escola","schule","Szkoła","школа","school","σχολείο"],"key::16":["education","educazione","éducation","educacion","Educação","Bildung","Edukacja","образование","opleiding","εκπαίδευση"],"key::17":["academy","accademia","académie","academia","academia","Akademie","akademie","академия","academie","ακαδημία"],"key::18":["public","pubblico","public","publique","publics","publiques","publico","publico","Öffentlichkeit","publiczny","публичный","publiek","publieke","δημόσιος","δημόσια","δημόσιο"],"key::19":["museum","museo","musée","mueso","museu","museum","muzeum","музей","museum","μουσείο"],"key::20":["group","gruppo","groupe","grupo","grupo","gruppe","grupa","группа","groep","ομάδα","όμιλος"],"key::21":["department","dipartimento","département","departamento","departamento","abteilung","departament","отдел","afdeling","τμήμα"],"key::22":["council","consiglio","conseil","Consejo","conselho","gesellschaft","rada","совет","raad","συμβούλιο"],"key::23":["library","biblioteca","bibliothèque","biblioteca","biblioteca","Bibliothek","biblioteka","библиотека","bibliotheek","βιβλιοθήκη"],"key::24":["ministry","ministero","ministère","ministerio","ministério","Ministerium","ministerstwo","министерство","ministerie","υπουργείο"],"key::25":["services","servizi","services","servicios","Serviços","Dienstleistungen","usługi","услуги","diensten","υπηρεσίες"],"key::26":["central","centrale","central","centrale","centrales","central","central","zentral","centralny","цетральный","centraal","κεντρικός","κεντρική","κεντρικό","κεντρικά"],"key::27":["general","generale","général","générale","généraux","générales","general","geral","general","Allgemeines","general","общий","algemeen","algemene","γενικός","γενική","γενικό","γενικά"],"key::28":["applied","applicati","appliqué","appliquée","appliqués","appliquées","aplicado","aplicada","angewendet","stosowany","прикладной","toegepast","toegepaste","εφαρμοσμένος","εφαρμοσμένη","εφαρμοσμένο","εφαρμοσμένα"],"key::29":["european","europee","europea","européen","européenne","européens","européennes","europeo","europeu","europäisch","europejski","европейский","Europees","Europese","ευρωπαϊκός","ευρωπαϊκή","ευρωπαϊκό","ευρωπαϊκά"],"key::30":["agency","agenzia","agence","agencia","agencia","agentur","agencja","агенция","agentschap","πρακτορείο"],"key::31":["laboratory","laboratorio","laboratoire","laboratorio","laboratorio","labor","laboratorium","лаборатория","laboratorium","εργαστήριο"],"key::32":["industry","industria","industrie","индустрия","industrie","βιομηχανία"],"key::33":["industrial","industriale","industriel","industrielle","industriels","industrielles","индустриальный","industrieel","βιομηχανικός","βιομηχανική","βιομηχανικό","βιομηχανικά","βιομηχανικές"],"key::34":["consortium","consorzio","consortium","консорциум","consortium","κοινοπραξία"],"key::35":["organization","organizzazione","organisation","organización","organização","organizacja","организация","organisatie","οργανισμός"],"key::36":["authority","autorità","autorité","авторитет","autoriteit"],"key::37":["federation","federazione","fédération","федерация","federatie","ομοσπονδία"],"key::38":["observatory","osservatorio","observatoire","обсерватория","observatorium","αστεροσκοπείο"],"key::39":["bureau","ufficio","bureau","офис","bureau","γραφείο"],"key::40":["company","impresa","compagnie","société","компания","bedrijf","εταιρία"],"key::41":["polytechnic","politecnico","polytechnique","политехника","polytechnisch","πολυτεχνείο","universita politecnica","polytechnic university","universidad politecnica","universitat politecnica","politechnika","politechniki","university technology","university science technology"],"key::42":["coalition","coalizione","coalition","коалиция","coalitie","συνασπισμός"],"key::43":["initiative","iniziativa","initiative","инициатива","initiatief","πρωτοβουλία"],"key::44":["academic","accademico","académique","universitaire","акадеческий academisch","ακαδημαϊκός","ακαδημαϊκή","ακαδημαϊκό","ακαδημαϊκές","ακαδημαϊκοί"],"key::45":["institution","istituzione","institution","институциональный","instelling","ινστιτούτο"],"key::46":["division","divisione","division","отделение","divisie","τμήμα"],"key::47":["committee","comitato","comité","комитет","commissie","επιτροπή"],"key::48":["promotion","promozione","продвижение","proothisis","forderung"],"key::49":["medical","medicine","clinical","medicina","clinici","médico","medicina","clínica","médico","medicina","clínica","medizinisch","Medizin","klinisch","medisch","geneeskunde","klinisch","ιατρικός","ιατρική","ιατρικό","ιατρικά","κλινικός","κλινική","κλινικό","κλινικά","tıbbi","tıp","klinik","orvosi","orvostudomány","klinikai","zdravniški","medicinski","klinični","meditsiini","kliinik","kliiniline"],"key::50":["technology","technological","tecnologia","tecnologie","tecnología","tecnológico","tecnologia","tecnológico","Technologie","technologisch","technologie","technologisch","τεχνολογία","τεχνολογικός","τεχνολογική","τεχνολογικό","teknoloji","teknolojik","technológia","technológiai","tehnologija","tehnološki","tehnoloogia","tehnoloogiline","technologii","technical","texniki","teknik"],"key::51":["science","scientific","scienza","scientifiche","scienze","ciencia","científico","ciência","científico","Wissenschaft","wissenschaftlich","wetenschap","wetenschappelijk","επιστήμη","επιστημονικός","επιστημονική","επιστημονικό","επιστημονικά","bilim","bilimsel","tudomány","tudományos","znanost","znanstveni","teadus","teaduslik",""],"key::52":["engineering","ingegneria","ingeniería","engenharia","Ingenieurwissenschaft","ingenieurswetenschappen","bouwkunde","μηχανικός","μηχανική","μηχανικό","mühendislik","mérnöki","Inženirstvo","inseneeria","inseneri",""],"key::53":["management","gestione","gestionale","gestionali","gestión","administración","gestão","administração","Verwaltung","management","διαχείριση","yönetim","menedzsment","vodstvo","upravljanje","management","juhtkond","juhtimine","haldus",""],"key::54":["energy","energia","energía","energia","Energie","energie","ενέργεια","enerji","energia","energija","energia",""],"key::55":["agricultural","agriculture","agricoltura","agricole","agrícola","agricultura","agrícola","agricultura","landwirtschaftlich","Landwirtschaft","landbouwkundig","landbouw","αγροτικός","αγροτική","αγροτικό","γεωργικός","γεωργική","γεωργικό","γεωργία","tarımsal","tarım","mezőgazdasági","mezőgazdaság","poljedelski","poljedelstvo","põllumajandus","põllumajanduslik",""],"key::56":["information","informazione","información","informação","Information","informatie","πληροφορία","bilgi","információ","informacija","informatsioon","informatycznych",""],"key::57":["social","sociali","social","social","Sozial","sociaal","maatschappelijk","κοινωνικός","κοινωνική","κοινωνικό","κοινωνικά","sosyal","szociális","družbeni","sotsiaal","sotsiaalne",""],"key::58":["environmental","ambiente","medioambiental","ambiente","medioambiente","meioambiente","Umwelt","milieu","milieuwetenschap","milieukunde","περιβαλλοντικός","περιβαλλοντική","περιβαλλοντικό","περιβαλλοντικά","çevre","környezeti","okoliški","keskonna",""],"key::59":["business","economia","economiche","economica","negocio","empresa","negócio","Unternehmen","bedrijf","bedrijfskunde","επιχείρηση","iş","üzleti","posel","ettevõte/äri",""],"key::60":["pharmaceuticals","pharmacy","farmacia","farmaceutica","farmacéutica","farmacia","farmacêutica","farmácia","Pharmazeutika","Arzneimittelkunde","farmaceutica","geneesmiddelen","apotheek","φαρμακευτικός","φαρμακευτική","φαρμακευτικό","φαρμακευτικά","φαρμακείο","ilaç","eczane","gyógyszerészeti","gyógyszertár","farmacevtika","lekarništvo","farmaatsia","farmatseutiline",""],"key::61":["healthcare","health services","salute","atenciónmédica","cuidadodelasalud","cuidadoscomasaúde","Gesundheitswesen","gezondheidszorg","ιατροφαρμακευτικήπερίθαλψη","sağlıkhizmeti","egészségügy","zdravstvo","tervishoid","tervishoiu",""],"key::62":["history","storia","historia","história","Geschichte","geschiedenis","geschiedkunde","ιστορία","tarih","történelem","zgodovina","ajalugu",""],"key::63":["materials","materiali","materia","materiales","materiais","materialen","υλικά","τεκμήρια","malzemeler","anyagok","materiali","materjalid","vahendid",""],"key::64":["economics","economia","economiche","economica","economía","economia","Wirtschaft","economie","οικονομικά","οικονομικέςεπιστήμες","ekonomi","közgazdaságtan","gospodarstvo","ekonomija","majanduslik","majandus",""],"key::65":["therapeutics","terapeutica","terapéutica","terapêutica","therapie","θεραπευτική","tedavibilimi","gyógykezelés","terapevtika","terapeutiline","ravi",""],"key::66":["oncology","oncologia","oncologico","oncología","oncologia","Onkologie","oncologie","ογκολογία","onkoloji","onkológia","onkologija","onkoloogia",""],"key::67":["natural","naturali","naturale","natural","natural","natürlich","natuurlijk","φυσικός","φυσική","φυσικό","φυσικά","doğal","természetes","naraven","loodus",""],"key::68":["educational","educazione","pedagogia","educacional","educativo","educacional","pädagogisch","educatief","εκπαιδευτικός","εκπαιδευτική","εκπαιδευτικό","εκπαιδευτικά","eğitimsel","oktatási","izobraževalen","haridus","hariduslik",""],"key::69":["biomedical","biomedica","biomédico","biomédico","biomedizinisch","biomedisch","βιοιατρικός","βιοιατρική","βιοιατρικό","βιοιατρικά","biyomedikal","orvosbiológiai","biomedicinski","biomeditsiiniline",""],"key::70":["veterinary","veterinaria","veterinarie","veterinaria","veterinária","tierärtzlich","veterinair","veeartsenijlkunde","κτηνιατρικός","κτηνιατρική","κτηνιατρικό","κτηνιατρικά","veteriner","állatorvosi","veterinar","veterinarski","veterinaaria",""],"key::71":["chemistry","chimica","química","química","Chemie","chemie","scheikunde","χημεία","kimya","kémia","kemija","keemia",""],"key::72":["security","sicurezza","seguridad","segurança","Sicherheit","veiligheid","ασφάλεια","güvenlik","biztonsági","varnost","turvalisus","julgeolek",""],"key::73":["biotechnology","biotecnologia","biotecnologie","biotecnología","biotecnologia","Biotechnologie","biotechnologie","βιοτεχνολογία","biyoteknoloji","biotechnológia","biotehnologija","biotehnoloogia",""],"key::74":["military","militare","militari","militar","militar","Militär","militair","leger","στρατιωτικός","στρατιωτική","στρατιωτικό","στρατιωτικά","askeri","katonai","vojaški","vojni","militaar","wojskowa",""],"key::75":["theological","teologia","teologico","teológico","tecnológica","theologisch","theologisch","θεολογικός","θεολογική","θεολογικό","θεολογικά","teolojik","technológiai","teološki","teoloogia","usuteadus","teoloogiline",""],"key::76":["electronics","elettronica","electrónica","eletrônicos","Elektronik","elektronica","ηλεκτρονική","elektronik","elektronika","elektronika","elektroonika",""],"key::77":["forestry","forestale","forestali","silvicultura","forestal","floresta","Forstwirtschaft","bosbouw","δασοκομία","δασολογία","ormancılık","erdészet","gozdarstvo","metsandus",""],"key::78":["maritime","marittima","marittime","marittimo","marítimo","marítimo","maritiem","ναυτικός","ναυτική","ναυτικό","ναυτικά","ναυτιλιακός","ναυτιλιακή","ναυτιλιακό","ναυτιλιακά","θαλάσσιος","θαλάσσια","θαλάσσιο","denizcilik","tengeri","morski","mere","merendus",""],"key::79":["sports","sport","deportes","esportes","Sport","sport","sportwetenschappen","άθληση","γυμναστικήδραστηριότητα","spor","sport","šport","sport","spordi",""],"key::80":["surgery","chirurgia","chirurgiche","cirugía","cirurgia","Chirurgie","chirurgie","heelkunde","εγχείρηση","επέμβαση","χειρουργικήεπέμβαση","cerrahi","sebészet","kirurgija","kirurgia",""],"key::81":["cultural","culturale","culturali","cultura","cultural","cultural","kulturell","cultureel","πολιτιστικός","πολιτιστική","πολιτιστικό","πολιτισμικός","πολιτισμική","πολιτισμικό","kültürel","kultúrális","kulturni","kultuuri","kultuuriline",""],"key::82":["computerscience","informatica","ordenador","computadora","informática","computación","cienciasdelacomputación","ciênciadacomputação","Computer","computer","υπολογιστής","ηλεκτρονικόςυπολογιστής","bilgisayar","számítógép","računalnik","arvuti",""],"key::83":["finance","financial","finanza","finanziarie","finanza","financiero","finanças","financeiro","Finanzen","finanziell","financiën","financieel","χρηματοοικονομικά","χρηματοδότηση","finanse","finansal","pénzügy","pénzügyi","finance","finančni","finants","finantsiline",""],"key::84":["communication","comunicazione","comuniciación","comunicação","Kommunikation","communication","επικοινωνία","iletişim","kommunikáció","komuniciranje","kommunikatsioon",""],"key::85":["justice","giustizia","justicia","justiça","Recht","Justiz","justitie","gerechtigheid","δικαιοσύνη","υπουργείοδικαιοσύνης","δίκαιο","adalet","igazságügy","pravo","õigus",""],"key::86":["aerospace","aerospaziale","aerospaziali","aeroespacio","aeroespaço","Luftfahrt","luchtvaart","ruimtevaart","αεροπορικός","αεροπορική","αεροπορικό","αεροναυπηγικός","αεροναυπηγική","αεροναυπηγικό","αεροναυπηγικά","havacılıkveuzay","légtér","zrakoplovstvo","atmosfäär","kosmos",""],"key::87":["dermatology","dermatologia","dermatología","dermatologia","Dermatologie","dermatologie","δρματολογία","dermatoloji","bőrgyógyászat","dermatológia","dermatologija","dermatoloogia",""],"key::88":["architecture","architettura","arquitectura","arquitetura","Architektur","architectuur","αρχιτεκτονική","mimarlık","építészet","arhitektura","arhitektuur",""],"key::89":["mathematics","matematica","matematiche","matemáticas","matemáticas","Mathematik","wiskunde","mathematica","μαθηματικά","matematik","matematika","matematika","matemaatika",""],"key::90":["language","lingue","linguistica","linguistiche","lenguaje","idioma","língua","idioma","Sprache","taal","taalkunde","γλώσσα","dil","nyelv","jezik","keel",""],"key::91":["neuroscience","neuroscienza","neurociencia","neurociência","Neurowissenschaft","neurowetenschappen","νευροεπιστήμη","nörobilim","idegtudomány","nevroznanost","neuroteadused",""],"key::92":["automation","automazione","automatización","automação","Automatisierung","automatisering","αυτοματοποίηση","otomasyon","automatizálás","avtomatizacija","automatiseeritud",""],"key::93":["pediatric","pediatria","pediatriche","pediatrico","pediátrico","pediatría","pediátrico","pediatria","pädiatrisch","pediatrische","παιδιατρική","pediatrik","gyermekgyógyászat","pediatrija","pediaatria",""],"key::94":["photonics","fotonica","fotoniche","fotónica","fotônica","Photonik","fotonica","φωτονική","fotonik","fotonika","fotonika","fotoonika",""],"key::95":["mechanics","meccanica","meccaniche","mecánica","mecânica","Mechanik","Maschinenbau","mechanica","werktuigkunde","μηχανικής","mekanik","gépészet","mehanika","mehaanika",""],"key::96":["psychiatrics","psichiatria","psichiatrica","psichiatriche","psiquiatría","psiquiatria","Psychiatrie","psychiatrie","ψυχιατρική","psikiyatrik","pszihiátria","psihiatrija","psühhaatria",""],"key::97":["psychology","fisiologia","psicología","psicologia","Psychologie","psychologie","ψυχολογία","psikoloji","pszihológia","psihologija","psühholoogia",""],"key::98":["automotive","industriaautomobilistica","industriadelautomóvil","automotriz","industriaautomotriz","automotivo","Automobilindustrie","autoindustrie","αυτοκίνητος","αυτοκίνητη","αυτοκίνητο","αυτοκινούμενος","αυτοκινούμενη","αυτοκινούμενο","αυτοκινητιστικός","αυτοκινητιστική","αυτοκινητιστικό","otomotiv","autóipari","samogiben","avtomobilskaindustrija","auto-",""],"key::99":["neurology","neurologia","neurologiche","neurología","neurologia","Neurologie","neurologie","zenuwleer","νευρολογία","nöroloji","neurológia","ideggyógyászat","nevrologija","neuroloogia",""],"key::100":["geology","geologia","geologiche","geología","geologia","Geologie","geologie","aardkunde","γεωλογία","jeoloji","geológia","földtudomány","geologija","geoloogia",""],"key::101":["microbiology","microbiologia","micro-biologia","microbiologiche","microbiología","microbiologia","Mikrobiologie","microbiologie","μικροβιολογία","mikrobiyoloji","mikrobiológia","mikrobiologija","mikrobioloogia",""],"key::102":["informatics","informatica","informática","informática","informatica",""],"key::103":["forschungsgemeinschaft","comunita ricerca","research community","research foundation","research association"],"key::104":["commerce","ticaret","ticarət","commercio","trade","handel","comercio"]}}}</value>
|
||
</property>
|
||
|
||
</configuration> |