Mariana Ferreira Dias

Abstract
Entity linking is an important task in medical natural language processing (NLP) for converting unstructured text into structured data for clinical analysis and semantic interoperability. However, in lower-resource languages, this task is challenging due to the limited availability of domain-specific resources. This paper explores a translation-based cross-lingual entity linking approach using GPT models, GPT-3.5 and GPT-4o, for zero-shot machine translation and entity linking with in-context learning. We evaluate our approach using a Portuguese-English parallel dataset of radiology abstracts. Our results show that chunk-level machine translation outperforms sentence-level translation. Moreover, our translation-based approach to cross-lingual entity linking of UMLS concepts outperformed the multilingual encoder method baseline. However, the in-context learning entity linking approach did not outperform a translation-based approach with a dictionary-based entity linking method. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Abstract
Linked data is used in various fields as a new way of structuring and connecting data. Cultural heritage institutions have been using linked data to improve archival descriptions and facilitate the discovery of information. Most archival records have digital representations of physical artifacts in the form of scanned images that are non-machine-readable. Optical Character Recognition (OCR) recognizes text in images and translates it into machine-encoded text. This article evaluates the impact of image processing methods and parameter tuning in OCR applied to typewritten cultural heritage documents. The approach uses a multi-objective problem formulation to minimize Levenshtein edit distance and maximize the number of words correctly identified with a non-dominated sorting genetic algorithm (NSGA-II) to tune the methods' parameters. Evaluation results show that parameterization by digital representation typology benefits the performance of image pre-processing algorithms in OCR. Furthermore, our findings suggest that employing image pre-processing algorithms in OCR might be more suitable for typologies where the text recognition task without pre-processing does not produce good results. In particular, Adaptive Thresholding, Bilateral Filter, and Opening are the best-performing algorithms for the theater plays' covers, letters, and overall dataset, respectively, and should be applied before OCR to improve its performance.

Abstract
Linked Data is used in various fields as a new way of structuring and connecting data. Cultural heritage institutions have been using linked data to improve archival descriptions and promote findability. The required detail in manual descriptions of cultural heritage objects can be taxing and time-consuming. Given this, in EPISA, a research project on this topic, we propose to use the contents of the digital representations associated with the objects to assist archivists in their description tasks. More specifically, to extract information from the digital representations useful for an initial ontology population that should be validated or edited by the archivist. We apply optical character recognition in an initial stage to convert the digital representation to a machine-readable format. We then use ontology-oriented programming to identify and instantiate ontology concepts using neural networks and contextual embeddings. © 2022 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0)