Danish Asylum Adjudication using Deep Neural Networks and Natural Language Processing
Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
TheDanishasylumadjudicationprocedure isatwo-tiered system,withtheImmigrationServicemakinginitialdeterminationsand theDanishRefugeeAppealsBoard(RAB)automaticallyappealingcases thatarerejected.Thisstudyaimstoemployadeepneuralnetwork(DNN)basedNaturalLanguageProcessing(NLP)pipeline topredictasylum decision-makingoutcomesusingadataset of over 15,515DanishasylumdecisionsprovidedbytheDanishRefugeeAppealsBoard(RAB) betweenJanuary1995andJanuary2021.Thisresearchseekstoimprove theperformanceandeffectivenessofdecision-makinginasylumcasesby addressingkeychallenges,suchasmodelingtheasylumdecision-making problemusingNLP-basedDNNsanddealingwithclassimbalanceissues. OurpreliminaryresultsindicatethatDNN-basedNLPpredictivemodels arecapableof learningmeaningful representationsofasylumcaseswith highprecisionandrecall,particularlywhenclassweightsareconsidered thanthebaselineDNNmodel.
Original language | English |
---|---|
Title of host publication | Danish Asylum Adjudication using Deep Neural Networks and Natural Language Processing |
Number of pages | 14 |
Publisher | Springer |
Pages | 1-13 |
Publication status | Accepted/In press - 2023 |
Links
- https://vbn.aau.dk/en/publications/danish-asylum-adjudication-using-deep-neural-networks-and-natural
Accepted author manuscript
ID: 377826428