implementation of new aggregation in the tree node processing

This commit is contained in:
miconis 2019-12-18 16:19:26 +01:00
parent 492049b8bc
commit 4af490221b
6 changed files with 0 additions and 495 deletions

View File

@ -1,40 +0,0 @@
{
"wf" : {
"threshold" : "0.99",
"dedupRun" : "001",
"entityType" : "person",
"orderField" : "fullname",
"queueMaxSize" : "2000",
"groupMaxSize" : "10",
"slidingWindowSize" : "200",
"rootBuilder" : [ "person" ],
"includeChildren" : "true"
},
"pace": {
"clustering": [
{"name": "personClustering", "fieldsCount": ["fullname"], "params": {}}
],
"necessaryConditions": [],
"decisionTree": {
"start": {"fieldsCount": [{"field":"pubID", "comparator":"exactMatch", "weight":1.0, "params":{}}], "threshold":1.0, "aggregation": "SUM", "positive":"NO_MATCH", "negative":"layer2", "undefined": "layer2", "countIfUndefined": "false"},
"layer2": {"fieldsCount": [{"field":"orcid", "comparator":"exactMatch", "weight":1.0, "params":{}}], "threshold":1.0, "aggregation": "SUM", "positive":"ORCID_MATCH", "negative":"NO_MATCH", "undefined": "layer3", "countIfUndefined": "false"},
"layer3": {"fieldsCount": [{"field":"firstname", "comparator":"similar", "weight":1.0, "params":{}}], "threshold":0.7, "aggregation": "SUM", "positive":"layer4", "negative":"NO_MATCH", "undefined": "layer4", "countIfUndefined": "false"},
"layer4": {"fieldsCount": [{"field":"coauthors", "comparator":"coauthorsMatch", "weight":1.0, "params":{"minCoauthors":6, "maxCoauthors": 200}}], "threshold":5.0, "aggregation": "SUM", "positive":"COAUTHORS_MATCH", "negative":"NO_MATCH", "undefined": "layer5", "countIfUndefined": "false"},
"layer5": {"fieldsCount": [{"field":"area", "comparator":"exactMatch", "weight":1.0, "params":{}}], "threshold":1.0, "aggregation": "SUM", "positive":"layer6", "negative":"NO_MATCH", "undefined": "NO_MATCH", "countIfUndefined": "false"},
"layer6": {"fieldsCount": [{"field":"topics", "comparator":"topicsMatch", "weight":1.0, "params":{}}], "threshold":0.7, "aggregation": "SUM", "positive":"TOPICS_MATCH", "negative":"NO_MATCH", "undefined": "NO_MATCH", "countIfUndefined": "false"}
},
"model": [
{"name": "fullname", "algo": "Null", "type": "String", "weight": "0", "countIfUndefined": "false", "path": "person/metadata/fullname"},
{"name": "firstname", "algo": "Null", "type": "String", "weight": "0", "countIfUndefined": "false", "path": "person/metadata/firstname"},
{"name": "lastname", "algo": "Null", "type": "String", "weight": "0", "countIfUndefined": "false", "path": "person/metadata/lastname"},
{"name": "coauthors", "algo": "Null", "type": "String", "weight": "0", "countIfUndefined": "false", "path": "person/metadata/coauthors"},
{"name": "orcid", "algo": "Null", "type": "String", "weight": "0", "countIfUndefined": "false", "path": "person/metadata/orcid"},
{"name": "topics", "algo": "Null", "type": "String", "weight": "0", "countIfUndefined": "false", "path": "person/metadata/topics"},
{"name": "pubID", "algo": "Null", "type": "String", "weight": "0", "countIfUndefined": "false", "path": "person/metadata/pubID"},
{"name": "pubDOI", "algo": "Null", "type": "String", "weight": "0", "countIfUndefined": "false", "path": "person/metadata/pubDOI"},
{"name": "rank", "algo": "Null", "type": "Int", "weight": "0", "countIfUndefined": "false", "path": "person/metadata/rank"},
{"name": "area", "algo": "Null", "type": "String", "weight": "0", "countIfUndefined": "false", "path": "person/metadata/area"}
],
"blacklists": {}
}
}

View File

@ -1,146 +0,0 @@
{
"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",
"maxIterations": "20"
},
"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} }
],
"decisionTree" : {
"start": {"fields": [{"field":"gridid", "comparator":"exactMatch", "weight":1.0, "countIfUndefined":"true", "params": {}}], "threshold":1.0, "aggregation": "MAX", "positive":"MATCH", "negative":"layer2", "undefined":"layer2", "ignoreUndefined": "true"},
"layer2": {"fields": [{"field":"websiteurl", "comparator":"domainExactMatch", "weight":1.0, "countIfUndefined":"true", "params" : {}}, {"field":"country", "comparator":"exactMatch", "weight":1.0, "countIfUndefined":"false", "params": {}}], "threshold":1.0, "aggregation": "MIN", "positive":"layer3", "negative":"NO_MATCH", "undefined":"layer3", "ignoreUndefined": "false"},
"layer3": {"fields": [{"field":"legalname", "comparator":"jaroWinklerNormalizedName", "weight":0.9, "countIfUndefined":"false", "params":{"windowSize" : 4, "threshold" : 0.7}}, {"field":"legalshortname", "comparator":"jaroWinklerNormalizedName", "weight":0.1, "countIfUndefined":"true", "params":{}}], "threshold": 0.9, "aggregation": "W_MEAN", "positive":"MATCH", "negative":"NO_MATCH", "undefined":"NO_MATCH", "ignoreUndefined": "true"}
},
"model" : [
{ "name" : "country", "algo" : "Null", "type" : "String", "weight" : "0", "ignoreMissing" : "false", "path" : ".organization.metadata.country.classid" },
{ "name" : "legalshortname", "algo" : "JaroWinklerNormalizedName", "type" : "String", "weight" : "0.1", "ignoreMissing" : "true", "path" : ".organization.metadata.legalshortname.value" },
{ "name" : "legalname", "algo" : "JaroWinklerNormalizedName", "type" : "String", "weight" : "0.9", "ignoreMissing" : "false", "path" : ".organization.metadata.legalname.value", "params" : {"windowSize" : 4, "threshold" : 0.7} },
{ "name" : "websiteurl", "algo" : "Null", "type" : "URL", "weight" : "0", "ignoreMissing" : "true", "path" : ".organization.metadata.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" : ["University of Turin"]
},
"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"]
}
}
}

View File

@ -1,40 +0,0 @@
{
"wf" : {
"threshold" : "0.99",
"dedupRun" : "001",
"entityType" : "result",
"subEntityType" : "resulttype",
"subEntityValue" : "software",
"orderField" : "title",
"queueMaxSize" : "2000",
"groupMaxSize" : "10",
"slidingWindowSize" : "200",
"rootBuilder" : [ "result", "resultProject_outcome_isProducedBy", "resultResult_publicationDataset_isRelatedTo", "resultResult_similarity_isAmongTopNSimilarDocuments", "resultResult_similarity_hasAmongTopNSimilarDocuments", "resultOrganization_affiliation_isAffiliatedWith", "resultResult_part_hasPart", "resultResult_part_isPartOf", "resultResult_supplement_isSupplementTo", "resultResult_supplement_isSupplementedBy", "resultResult_version_isVersionOf" ],
"includeChildren" : "true"
},
"pace" : {
"clustering" : [
{ "name" : "ngrampairs", "fieldsCount" : [ "title" ], "params" : { "max" : "1", "ngramLen" : "3"} },
{ "name" : "suffixprefix", "fieldsCount" : [ "title" ], "params" : { "max" : "1", "len" : "3" } },
{ "name" : "lowercase", "fieldsCount" : [ "doi" ], "params" : { } },
{ "name" : "urlclustering", "fieldsCount": [ "url" ], "params" : { } }
],
"sufficientConditions" : [
{ "name" : "doiExactMatch", "fieldsCount": [ "doi" ] },
{ "name" : "exactMatch", "fieldsCount" : [ "url", "documentationUrl" ] }
],
"necessaryConditions" : [
{ "name" : "exactMatch", "fieldsCount" : ["resulttype"] }
],
"model" : [
{ "name" : "doi", "algo" : "Null", "type" : "String", "weight" : "0.0", "countIfUndefined" : "true", "path" : "pid[qualifier#classid = {doi}]/value" },
{ "name" : "title", "algo" : "LevensteinTitleIgnoreVersion", "type" : "String", "weight" : "1.0", "countIfUndefined" : "false", "path" : "result/metadata/title[qualifier#classid = {main title}]/value", "length" : 250, "size" : 5 },
{ "name" : "url", "algo" : "Null", "type" : "String", "weight" : "0.0", "countIfUndefined" : "true", "path" : "result/instance/url" },
{ "name" : "resulttype", "algo" : "Null", "type" : "String", "weight" : "0.0", "countIfUndefined" : "false", "path" : "result/metadata/resulttype/classid" },
{ "name" : "documentationUrl", "algo" : "Null", "type" : "String", "weight" : "0.0", "countIfUndefined" : "false", "path" : "result/metadata/documentationUrl/value" }
],
"blacklists" : {
}
}
}

View File

@ -1,269 +0,0 @@
{
"journal": {
"name": "",
"issnPrinted": "",
"issnOnline": "",
"issnLinking": "",
"ep": "",
"iss": "",
"sp": "",
"vol": "",
"edition": "",
"conferenceplace": "",
"conferencedate": "",
"dataInfo": {
"invisible": false,
"inferred": false,
"deletedbyinference": false,
"inferenceprovenance": "",
"provenanceaction": {
"classid": "",
"classname": "",
"schemeid": "",
"schemename": ""
}
}
},
"author": [
{
"fullname": "Clingerman Daniel J.",
"name": "",
"surname": "",
"rank": 1,
"pid": [],
"affiliation": []
},
{
"fullname": "Morris William",
"name": "",
"surname": "",
"rank": 2,
"pid": [],
"affiliation": []
},
{
"fullname": "Sarjeant Amy A.",
"name": "",
"surname": "",
"rank": 3,
"pid": [],
"affiliation": []
},
{
"fullname": "Farha Omar K.",
"name": "",
"surname": "",
"rank": 4,
"pid": [],
"affiliation": []
},
{
"fullname": "Mondloch Joseph E.",
"name": "",
"surname": "",
"rank": 5,
"pid": [],
"affiliation": []
},
{
"fullname": "Hupp Joseph T.",
"name": "",
"surname": "",
"rank": 6,
"pid": [],
"affiliation": []
},
{
"fullname": "Stern Charlotte",
"name": "",
"surname": "",
"rank": 7,
"pid": [],
"affiliation": []
},
{
"fullname": "Kennedy Robert D.",
"name": "",
"surname": "",
"rank": 8,
"pid": [],
"affiliation": []
},
{
"fullname": "Mirkin Chad A.",
"name": "",
"surname": "",
"rank": 9,
"pid": [],
"affiliation": []
}
],
"resulttype": {
"classid": "publication",
"classname": "publication",
"schemeid": "dnet:result_typologies",
"schemename": "dnet:result_typologies"
},
"language": {
"classid": "und",
"classname": "Undetermined",
"schemeid": "dent:languages",
"schemename": "dent:languages"
},
"country": [],
"subject": [],
"title": [
{
"value": "Stabilization of a highly porous metal\u2013organic framework utilizing a carborane-based linker",
"qualifier": {
"classid": "main title",
"classname": "main title",
"schemeid": "dnet:dataCite_title",
"schemename": "dnet:dataCite_title"
},
"dataInfo": {
"invisible": false,
"inferred": false,
"deletedbyinference": false,
"inferenceprovenance": "",
"provenanceaction": {
"classid": "",
"classname": "",
"schemeid": "",
"schemename": ""
}
}
}
],
"relevantdate": [],
"description": [],
"dateofacceptance": {
"value": "",
"dataInfo": {
"invisible": false,
"inferred": false,
"deletedbyinference": false,
"inferenceprovenance": "",
"provenanceaction": {
"classid": "",
"classname": "",
"schemeid": "",
"schemename": ""
}
}
},
"publisher": {
"value": "Royal Society of Chemistry (RSC)",
"dataInfo": {
"invisible": false,
"inferred": false,
"deletedbyinference": false,
"inferenceprovenance": "",
"provenanceaction": {
"classid": "",
"classname": "",
"schemeid": "",
"schemename": ""
}
}
},
"embargoenddate": {
"value": "",
"dataInfo": {
"invisible": false,
"inferred": false,
"deletedbyinference": false,
"inferenceprovenance": "",
"provenanceaction": {
"classid": "",
"classname": "",
"schemeid": "",
"schemename": ""
}
}
},
"source": [],
"fulltext": [],
"format": [],
"contributor": [],
"resourcetype": {
"classid": "",
"classname": "",
"schemeid": "",
"schemename": ""
},
"coverage": [],
"refereed": {
"value": "",
"dataInfo": {
"invisible": false,
"inferred": false,
"deletedbyinference": false,
"inferenceprovenance": "",
"provenanceaction": {
"classid": "",
"classname": "",
"schemeid": "",
"schemename": ""
}
}
},
"context": [],
"id": "50|scholexplore::cf30b31310d816cccd28b514a12ea4a0",
"originalId": [],
"collectedfrom": [
{
"key": "10|openaire____::e034d6a11054f5ade9221ebac484e864",
"value": "scholExplorer",
"dataInfo": {
"invisible": false,
"inferred": false,
"deletedbyinference": false,
"inferenceprovenance": "",
"provenanceaction": {
"classid": "",
"classname": "",
"schemeid": "",
"schemename": ""
}
}
}
],
"pid": [
{
"value": "10.1039/c4cc09212k",
"qualifier": {
"classid": "doi",
"classname": "doi",
"schemeid": "dnet:pid_types",
"schemename": "dnet:pid_types"
},
"dataInfo": {
"invisible": false,
"inferred": false,
"deletedbyinference": false,
"inferenceprovenance": "",
"provenanceaction": {
"classid": "",
"classname": "",
"schemeid": "",
"schemename": ""
}
}
}
],
"extraInfo": [],
"dataInfo": {
"invisible": false,
"inferred": true,
"deletedbyinference": true,
"inferenceprovenance": "dedup-similarity-result-levenstein",
"provenanceaction": {
"classid": "sysimport:actionset",
"classname": "sysimport:actionset",
"schemeid": "dnet:provenanceActions",
"schemename": "dnet:provenanceActions"
}
},
"lastupdatetimestamp": 0
}