Cognitive Technologies Georg Rehm Editor European Language Grid A Language Technology Platform for Multilingual Europe Cognitive Technologies Editor-in-Chief Daniel Sonntag, German Research Center for AI, DFKI, Saarbrücken, Saarland, Germany Titles in this series now included in the Thomson Reuters Book Citation Index and Scopus! The Cognitive Technologies (CT) series is committed to the timely publishing of high-quality manuscripts that promote the development of cognitive technologies and systems on the basis of artificial intelligence, image processing and understanding, natural language processing, machine learning and human-computer interaction. It brings together the latest developments in all areas of this multidisciplinary topic, ranging from theories and algorithms to various important applications. The intended readership includes research students and researchers in computer science, computer engineering, cognitive science, electrical engineering, data science and related fields seeking a convenient way to track the latest findings on the foundations, methodologies and key applications of cognitive technologies. The series provides a publishing and communication platform for all cognitive technologies topics, including but not limited to these most recent examples: Interactive machine learning, interactive deep learning, machine teaching Explainability (XAI), transparency, robustness of AI and trustworthy AI Knowledge representation, automated reasoning, multiagent systems Common sense modelling, context-based interpretation, hybrid cognitive technologies Human-centered design, socio-technical systems, human-robot interaction, cognitive robotics Learning with small datasets, never-ending learning, metacognition and introspection Intelligent decision support systems, prediction systems and warning systems Special transfer topics such as CT for computational sustainability, CT in business applications and CT in mobile robotic systems The series includes monographs, introductory and advanced textbooks, state-of- the-art collections, and handbooks. In addition, it supports publishing in Open Access mode. Georg Rehm Editor European Language Grid A Language Technology Platform for Multilingual Europe Editor Georg Rehm Deutsches Forschungszentrum für Künstliche Intelligenz GmbH (DFKI) Berlin, Germany The European Language Grid has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement no. 825627. ISSN 1611-2482 ISSN 2197-6635 (electronic) Cognitive Technologies ISBN 978-3-031-17257-1 ISBN 978-3-031-17258-8 (eBook) https://doi.org/10.1007/978-3-031-17258-8 © The Editor(s) (if applicable) and The Author(s) 2023. This book is an open access publication. Open Access This book is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made. The images or other third party material in this book are included in the book's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the book's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland Foreword IwasproudtohavetheopportunitytopresentmyreportonLanguageEqualityin theDigitalAgetotheEuropeanParliamentin2018andevenproudertoseetheover- whelmingsupportitreceived.ItwasoneofmyfinalachievementsasaMemberof theEuropeanParliamentandIamdelightedthatitcontributedtothegroundbreak- ingworkbeingdoneontheEuropeanLanguageGridproject.Despiteitnotbeinga legislativereport,thelevelofcross-partysupportitreceivedmeantitsrecommenda- tionscouldnotbeignored. WhenIfirstproposedareportonlanguageequalityinthedigitalagetotheEu- ropean Parliament’s Culture and Education Committee it provoked a great deal of interest, as it did in the Industry Committee. This was due to the clear language inequalityinEuropebutalsotothehugeopportunitiesitpresentedforthedigitalin- dustries.Asbothcommitteeslaidclaimtothereport,therewassomedebatebefore itwasresolvedthatitwouldbeaCultureCommitteereportwithawrittenopinion fromtheIndustryCommittee.Thelatter’sparticipationstrengthenedthereportand itsimpact.Itwidenedthescopetoemphasisetheimportanceoftheroleofprivate companiesalongsidepublicbodiesandoffacilitatingcrossbordertradeintheDigi- talSingleMarket. ItwasclearfromtheearlydaysthattheEuropeanCommissionwaskeentosup- portthereportandtaketheproposalsforward.AsaspokespersonfortheCommis- sioner stated in a conference I organised in parliament in September 2018, “You areneversowealthyaswhenyoucanspeakinyourownlanguage”.TheEuropean Language Equality project is currently developing a roadmap to achieve language equalityby2030,whichwillbepresentedtotheEuropeanInstitutionslaterthisyear. Minoritylanguagesinparticularhavemosttolosebutalsomosttogainfromthe digitalage,giventherightsupport.Culturalandlinguisticdiversitydependslargely onthetechnologicalresourcesavailabletoalllanguages. ItwasareportbytheEUPanelfortheFutureofScienceandTechnology,STOA, that sparked the idea of a parliamentary report. STOA highlighted the social and economicconsequencesoflanguagebarriersandthewideningofthetechnological gap.AssomeonewhohadlongcampaignedforequalstatusfortheWelshlanguage, IwasinspiredbythepotentialthatamajorEUprojectcouldoffer. v vi Foreword Even though the dominance of a few well-resourced languages in the digital worldwasobvious,theimpactofthisonotherlanguageshadnotbeenadequately explored.Whenthediscussionbegan,theinterestgrew.Theincreaseintechnology presented new threats and new opportunities. This was an issue which literally af- fected everyone, and most notably children growing up in this digital world. The roleofeducationiscrucialinteachingandunderstandinglanguagetechnologiesbut alsoinraisingawarenessofcareeropportunitiesinthisindustryacrossEurope. TheEuropeanUnionitself,ofcourse,couldcouldplayamajorrole.Theinstitu- tionalframeworkfortheprovisionoflanguagetechnologycouldbeimprovedcon- siderably. I believed that this was such a crucial issue that it deserved the specific allocationoftheportfoliotoaEuropeanCommissioner.Thisdidnotmaterialisein the appointment of a new Commission following the European elections in 2019, butIbelievetheproposalshouldbemaintainedandcouldbeadoptedinfuture. ThestrongsupportgiventomyreportbytheEuropeanParliamentwasanindica- tionofsupportfortheexcitinglanguageequalityworkthatistakingplacenow.The reportproposedadedicatedfundingprogrammeforresearch,developmentandinno- vationinlanguagetechnologieswiththeaimofclosingthegapbetweenEuropean languages. Thissuggestionwasadirectresultofseeingtheexistingresearchbeingdonein many countries. Identifying the problem went hand in hand with discovering that thereweremanyindividualsandorganisationsalreadyaddressingitandworkingto overcome it. They had the information and expertise but needed far more support andahigherprofile.ItwasclearthattheEUcouldbecomeatrailblazerinresearch ondigitallanguagetechnology,giventhepoliticalwill. Asapolitician,therightsofminoritylanguageslikemyown,Welsh,wereatthe heartofmyworkforjusticeandequality.Forme,languagewasnotmerelyameans of communication but central to our culture and identity. The EU claims equality indiversitybutwhenitcametolanguageequalityitfellfarshort.Soinmyroleas a Member of the European Parliament I saw an opportunity to help correct this. I couldplaymyroleinparliamentbuttoensurethereportwaseffectiveinachieving itsaimitneededtheinputoftheexperts,thepractitionersandthepioneersinthis fieldofworktoensurethatitwasaccurateandinformed. IneverfailtobeinspiredbytheirworkandtheirdedicationandIrepeatmythanks toallthosewhocontributedtothesuccessoftheLanguageEqualityintheDigital AgereportandtotheremarkableEuropeanLanguageGridprojectwhichestablished theprimaryplatformfor“languagetechnologiesforEuropebuiltinEurope”. RhonddaValley,April2022 JillEvans Preface Theoriginsofthisbookdatebackto2012.Backthen,undertheumbrellaoftheEU NetworkofExcellenceMETA-NET,wepreparedtherecommendationsandpriority researchthemesspecifiedinthefirstStrategicResearchAgenda(SRA)fortheEu- ropeanLanguageTechnology(LT)fieldinacomplex,community-drivenprocess.1 WhiletheEuropeanLTcommunityisquiteextensive,withhundredsofcommercial andacademicorganisationsworkingonalargeandheterogeneoussetoftechnolo- gies,itisalsoextremelyfragmentedwithmanycommunitymembersoperatingonly in narrow niches and limited regions, on very specific topics and quite often only takingintoaccountoneortworegionallyconfinedlanguages.ThroughtheMETA- NETSRAprocess,wehavebeenabletoidentifythecommunity’sneedforajoint technologyplatformthatbrings theEuropean LTcommunitytogether,thatfosters collaborationandsynergies,thatactsasamarketplaceanddeploymentplatform,that functionslikethe“yellowpages”oftheEuropeanLTcommunityandthroughwhich essentiallyallEuropeanresources,corpora,datasetsandgrammarsaswellastools, servicesandsourcecodecanbediscoveredandactuallyused,straightfromtheplat- formitself.Backin2013,inthepublishedMETA-NETSRA,wecalledthisconcept theEuropeanServicePlatformforLanguageTechnologies.TheSRAdocumentonly contained a rather coarse-grained description of this ambitious technology vision, whichhasbeendemanded,foranumberofdifferentreasons,byanoverwhelming majorityofthemembersoftheLTcommunity. Lateron,inthethreeStrategicResearchandInnovationAgendaspreparedunder theumbrellaoftheEUprojectCRACKER(CrackingtheLanguageBarrier;2015- 2017),werefinedthenotionoftheEuropeanLTServicePlatformandweextended the possible use cases and a large number of LT-driven applications, primarily fo- cusingthemultilingualdigitalsinglemarket.Furtherboostedbythescientificbreak- throughsproducedintheareaofArtificialIntelligence,MachineLearningandDeep Learning,earlyonappliedtoLTapplicationssuchasMachineTranslation,notonly thetopicofLanguageTechnologybutalsothevisionofajointEuropeanLanguage TechnologyPlatformbecamemoreandmorerelevant.Thetopicwasmentionedin 1http://www.meta-net.eu/sra vii viii Preface aprominentwayintheSTOAstudyLanguageequalityinthedigitalage2,commis- sioned by the European Parliament and also in a European Parliament resolution3 withthesametitle,adoptedbytheEuropeanParliamentinalandslidevotein2018 (cf.JillEvans’foreword). Roughlyatthesametime,inlate2017,westartedpreparingaprojectproposal for the Horizon 2020 ICT call, topic ICT-29a), European Language Grid, which fortunatelyreflectedthevastinterestwithinthecommunityinsuchaplatform.After various unsuccessful attempts at coming up with a good title for the proposal, we decidedtousethetitleoftheactualcallbecauseitfitperfectly.Havingpassedthe evaluation with a positive result, the project started in January 2019. We had an enthusiastickick-offmeeting,excitinghackathonsanddevelopedthefirstprototype oftheplatforminafastandagileway.ItwasfirstpresentedtothepublicatMETA- FORUM2019,whichtookplaceinBrusselsinOctoberthatyear. Onlyafewweekslater,theglobalSARS-CoV-2pandemichit.Thewholeworld wasaffectedandsowasourprojectplan.Wewereunabletohaveface-to-faceproject meetings or additional hackathons, we were unable to organise any on-site work- shops with our 32 ELG National Competence Centres as part of the “ELG Euro- peanRoadshow”.Allmeetings,includingourannualMETA-FORUMconferences in2020and2021,hadtogovirtual,whichwasnewtousatfirstandquicklybecame thenewnormal.Recently,inearlyJune2022,wehadourlastMETA-FORUMcon- ferenceundertheumbrellaoftheELGEUproject.META-FORUM2022wentback, atleastpartially,totheold normalwithapprox.100participantsintheconference centreinBrusselsandhundredsmoreparticipatingremotely. ItwasnothingbutapleasuretoactasCoordinatoroftheEuropeanLanguageGrid projectandtoworktogetherwithsuchastronganddedicatedteam.Ouroriginalplan inthisInnovationActionwasalreadyquiteambitiousyetwemanagedtoexceedour jointexpectationsintermsofthetechnologyplatformanditsfeatures,intermsofthe servicesandresourcesdeveloped,collectedandingestedintotheplatform,interms oftheacceptanceandfeedbackbythecommunityandalsointermsofthevarious collaborations we conducted with other projects. Many of the features envisioned for the European Service Platform for Language Technologies in 2013 are in fact nowfinallyavailableintheEuropeanLanguageGrid,whichis,byalargemargin, the biggest all-purpose Language Technology platform on the planet covering the wholebreadthandtechnologyspectrumofthefield. All of the activities and results produced by the nine partners of the ELG con- sortium during the project’s runtime are described in this book in detail. I would like to thank all consortium partners and team members for their extremely hard and dedicated work towards our common goal of developing and establishing the ELGplatform,communityandmarketplace.Inaddition,Iwouldliketothankthe 15selectedpilotprojectsfortheirinnovativeproposalsandthemorethan200organ- isationswhoappliedforfundingthroughoneofourpilotprojects.Thanksarealso duetotheprojectsELGcollaboratedwith,especially,in2021/2022,theEuropean 2https://www.europarl.europa.eu/stoa/en/document/EPRS_STU(2017)598621 3https://www.europarl.europa.eu/doceo/document/TA-8-2018-0332_EN.html Preface ix LanguageEqualityproject,theresultsofwhichwillalsobedocumentedintheform ofabookinthesameseries,butalsootherssuchasBergamot,COMPRISE,ELITR, EMBEDDIA,Gourmet,Prêt-à-LLOD,AI4EU,HumanEAINet,VISION,TAILOR, WeVerify,NTEU,MicroservicesatyourService,MAPA,QURATOR,PANQURA, SPEAKERandmanyothers. ThisbookisthedefinitivedocumentationoftheEUprojectEuropeanLanguage Grid.4 IwouldliketothankallcolleaguesfromtheELGconsortiumandalsofrom the ELG pilot projects wholeheartedly for the chapters they contributed, without whichthisbookwouldnothavebeenpossible. Whilethisbookcanonlycovertheresultsachievedduringtheproject’sruntime (January2019untilJune2022),theELGinitiativewillcontinue.Inthesecondhalfof 2022wewillestablishalegalentitythatwilltakeovermaintenanceandoperationof theplatform.WehopethatELGwillserveitsmanypurposesand,amongothers,ad- dressthestarkcommunityfragmentationandcontributetodigitallanguageequality inEurope,functioningindeedasonejointumbrellaplatformforthewholeEuropean LTcommunity.Furthermore,whilenoneofthesecanbeconsideredadirectfollow- up just yet, in a few projects (including OpenGPT-X, NFDI4DataScience and AI aswellastheEUprojectsDataBri-XandSciLake)wewillhavetheopportunityto continueourworkwithandontheELGplatform. Berlin,July2022 GeorgRehm Acknowledgements TheEuropeanLanguageGridEUprojecthasreceivedfundingfromtheEuro- peanUnion’sHorizon2020researchandinnovationprogrammeundergrantagreementno.825627. 4 https://european-language-grid.readthedocs.ioprovidesmoredetailswithregardtotechnical aspectsoftheELGplatform.Theonlinedocumentationisactivelymaintainedandkeptuptodate.