Romanian Named Entity Recognition in the Legal domain (LegalNERo)

LegalNERo is a manually annotated corpus for named entity recognition in the Romanian legal domain. <br> It provides gold annotations for organizations, locations, persons, time and legal resources mentioned in legal documents (legal references).Starting with version 4, the legal references were annotated using fine-grained legal document types: Law. Ordinance, Publication, Decree, Decision, Treaty, Report, Order, Regulation, Directive, EmergencyOrdinance, Norm, Convention, Code, and Other.<br> Additionally it offers GEONAMES codes for the named entities annotated as location (where a link could be established). The LegalNERo corpus is available in different formats: span-based, token-based and RDF. <br> The Linguistic Linked Open Data (LLOD) version is provided in RDF-Turtle format. CONLLUP files conform to the CoNLL-U Plus format https://universaldependencies.org/ext-format.html .<br> Part-of-speech tagging was realized using UDPIPE. <br> Named entity annotations are placed in the column "RELATE:NE" (the 11th column) as defined in the "global.columns" metadata field.<br> Similarly GEONAMES references are in the column "RELATE:GEONAMES" (the 12th column, last).<br> Automatic processing was performed through the RELATE platform (https://relate.racai.ro). ANN files conform to BRAT format (https://brat.nlplab.org/).<br> <br> The archive contains: - ann_LEGAL_PER_LOC_ORG_TIME_overlap <br> Folder in which all the files are in .ann format and contains annotations of: legal references, persons, locations, organizations and time. <br> Overlapping annotations of organizations and time entities inside legal references were allowed. - ann_FGLEGAL_PER_LOC_ORG_TIME_overlap <br> Folder (corresponding to the above entry) in which all the files are in .ann format and contains annotations of: fine-grained legal references, persons, locations, organizations and time. <br> Overlapping annotations of organizations and time entities inside legal references were allowed. - ann_LEGAL_PER_LOC_ORG_TIME <br> Folder in which all the files are in .ann format and contains annotations of: legal references, persons, locations, organizations and time. <br> Overlapping annotations were not allowed and only the longest named entities were annotated. - ann_FGLEGAL_PER_LOC_ORG_TIME <br> Folder (corresponding to the above entry) in which all the files are in .ann format and contains annotations of: fine-grained legal references, persons, locations, organizations and time. <br> Overlapping annotations were not allowed and only the longest named entities were annotated. - ann_PER_LOC_ORG_TIME <br> Folder in which all the files are in .ann format and contains annotations of: persons, locations, organizations and time. <br> There are no overlapping annotations. - conllup_LEGAL_PER_LOC_ORG_TIME <br> Folder in which all the files are in .conllup format and contains annotations of: legal references, persons, locations, organizations and time. <br> Overlapping annotations were not allowed and only the longest named entities were annotated. <br> The annotation of these files was enhanced with GEONAMES codes (where linking was possible). - conllup_FGLEGAL_PER_LOC_ORG_TIME <br> Folder (corresponding to the above entry) in which all the files are in .conllup format and contains annotations of: fine-grained legal references, persons, locations, organizations and time. <br> Overlapping annotations were not allowed and only the longest named entities were annotated. <br> The annotation of these files was enhanced with GEONAMES codes (where linking was possible). - conllup_PER_LOC_ORG_TIME <br> Folder in which all the files are in .conllup format and contains annotations of: persons, locations, organizations and time. <br> Overlapping annotations were not allowed and only the longest named entities were annotated. <br> The annotation of these files was enhanced with GEONAMES codes (where linking was possible). - rdf <br> Folder containing the corpus in RDF-Turtle format.<br> All the annotations are available here in both span and token format. - text <br> Folder containing the raw texts. - splits_FGLEGAL_PER_LOC_ORG_TIME.tsv<br> This is a proposed split of the documents for training a NER system using the fine-grained entity classes. <br> The split was created randomly, while trying to ensure 15% of each entity type for validation, 15% for testing and 70% for training.<br> <strong>NER System</strong> A NER model generated using the LegalNERo corpus can be used online in the RELATE platform: https://relate.racai.ro/index.php?path=ner/demo This system was described in: Păiș, Vasile and Mitrofan, Maria and Gasan, Carol Luca and Coneschi, Vlad and Ianov, Alexandru. Named Entity Recognition in the Romanian Legal Domain. In Proceedings of the Natural Legal Language Processing Workshop 2021. Association for Computational Linguistics, Punta Cana, Dominican Republic, pp. 9--18, nov 2021 <br> <strong>LICENSING</strong> This work is provided under the license CC BY-NC-ND 4.0 (Attribution-NonCommercial-NoDerivatives 4.0 International).<br> The license can be viewed online here: https://creativecommons.org/licenses/by-nc-nd/4.0/ <br> and the full text here: https://creativecommons.org/licenses/by-nc-nd/4.0/legalcode . <br> <strong>CONTACT</strong> Research Institute for Artificial Intelligence "Mihai Draganescu", Romanian Academy<br> Web: http://www.racai.ro <br> Contact emails: [e-mail] , [e-mail]