Making Sense of the Ways We Make Sense of the Past: The Periodo Project

Rabinowitz, Adam; Shaw, Ryan; Buchanan, Sarah; Golden, Patrick; Kansa, Eric

doi:10.1111/j.2041-5370.2016.12037.x

Cited by 19 publications

(13 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…It uses regular expressions to convert dates (e.g. '100 BCE', 'start of the 9th century') and an extended and customised version of the PeriodO time period gazetteer [40] to translate Time Periods (e.g. 'Bronze Age', 'Medieval period').…”

Section: Entity-driven Document Searchmentioning

confidence: 99%

Can BERT Dig It? Named Entity Recognition for Information Retrieval in the Archaeology Domain

Brandsen

Verberne

Lambers

et al. 2022

J. Comput. Cult. Herit.

View full text Add to dashboard Cite

The amount of archaeological literature is growing rapidly. Until recently, these data were only accessible through metadata search. We implemented a text retrieval engine for a large archaeological text collection (∼ 658 Million words). In archaeological IR, domain-specific entities such as locations, time periods, and artefacts, play a central role. This motivated the development of a named entity recognition (NER) model to annotate the full collection with archaeological named entities. In this paper, we present ArcheoBERTje, a BERT model pre-trained on Dutch archaeological texts. We compare the model’s quality and output on a Named Entity Recognition task to a generic multilingual model and a generic Dutch model. We also investigate ensemble methods for combining multiple BERT models, and combining the best BERT model with a domain thesaurus using Conditional Random Fields (CRF). We find that ArcheoBERTje outperforms both the multilingual and Dutch model significantly with a smaller standard deviation between runs, reaching an average F1 score of 0.735. The model also outperforms ensemble methods combining the three models. Combining ArcheoBERTje predictions and explicit domain knowledge from the thesaurus did not increase the F1 score. We quantitatively and qualitatively analyse the differences between the vocabulary and output of the BERT models on the full collection and provide some valuable insights in the effect of fine-tuning for specific domains. Our results indicate that for a highly specific text domain such as archaeology, further pre-training on domain-specific data increases the model’s quality on NER by a much larger margin than shown for other domains in the literature, and that domain-specific pre-training makes the addition of domain knowledge from a thesaurus unnecessary.

show abstract

Section: Entity-driven Document Searchmentioning

confidence: 99%

Can BERT Dig It? Named Entity Recognition for Information Retrieval in the Archaeology Domain

Brandsen

Verberne

Lambers

et al. 2022

J. Comput. Cult. Herit.

View full text Add to dashboard Cite

show abstract

“…Una de las particularidades de la arqueología es el uso de la dimensión temporal. Aunque no es exclusiva de este tipo de estudios, a lo largo de la evolución de la disciplina se han generado modelos explicativos del desarrollo de las poblaciones humanas sobre la base de constructos teóricos tales como etapas y períodos (Rabinowitz et al, 2016;Aguilar, 2019;Izeta y Aguilar, 2020). Basándose en la diversidad de periodificaciones a escala global, se propuso su sistematización a partir del concepto de nanopublicación (Golden y Shaw, 2015), lo que ha producido, a la fecha, 6423 periodos.…”

Section: Periodo Nomenclador De Periodos Históricos Y Arqueológicosunclassified

Sentando las bases para una Arqueología Digital en Argentina. El rol de las infraestructuras digitales para la investigación

Izeta

Prado

Cattáneo

2021

View full text Add to dashboard Cite

En el marco de las políticas públicas relacionadas con el desarrollo de la Ciencia y Técnica en la Argentina es cada vez más evidente una orientación dirigida a la implementación de un modelo de ciencia abierta basado en la potencialidad ofrecida por los medios digitales. La arqueología no ha sido ajena a esto y consciente o inconscientemente se ha volcado hacia una práctica mediada por “lo digital”. Para este modelo la accesibilidad a los datos primarios y la construcción de infraestructuras digitales para la investigación es prioritario. Siguiendo esta idea se implementó el Repositorio Digital Institucional Suquía centrado en datos arqueológicos. En esta infraestructura digital se incluyen diversos tipos de información con una característica común: su difícil acceso, tanto físico como virtual, en su formato original. Bases de datos, informes, imágenes de excavaciones, de objetos y colecciones arqueológicas, son algunos de los tipos contenidos en el repositorio. Este nuevo tipo de infraestructura se propone como un elemento de cambio en cómo se piensa y realiza la práctica arqueológica, ya que permitirá el uso y reúso de datos generados por financiamiento público tanto para la investigación arqueológica, la gestión del patrimonio y la divulgación científica, como para interesados en la temática.

show abstract

“…It uses regular expressions to convert dates (e.g. '100 BCE', 'start of the 9th century') and an extended and customised version of the PeriodO time period gazetteer [38] to translate Time Periods (e.g. 'Bronze Age', 'Medieval period').…”

Section: Entity-driven Document Searchmentioning

confidence: 99%

Can BERT Dig It? -- Named Entity Recognition for Information Retrieval in the Archaeology Domain

Brandsen¹,

Verberne²,

Lambers³

et al. 2021

Preprint

View full text Add to dashboard Cite

The amount of archaeological literature is growing rapidly. Until recently, these data were only accessible through metadata search. We implemented a text retrieval engine for a large archaeological text collection (∼ 658 Million words). In archaeological IR, domain-specific entities such as locations, time periods, and artefacts, play a central role. This motivated the development of a named entity recognition (NER) model to annotate the full collection with archaeological named entities. In this paper, we present ArcheoBERTje, a BERT model pre-trained on Dutch archaeological texts. We compare the model's quality and output on a Named Entity Recognition task to a generic multilingual model and a generic Dutch model. We also investigate ensemble methods for combining multiple BERT models, and combining the best BERT model with a domain thesaurus using Conditional Random Fields (CRF). We find that ArcheoBERTje outperforms both the multilingual and Dutch model significantly with a smaller standard deviation between runs, reaching an average F1 score of 0.735. The model also outperforms ensemble methods combining the three models. Combining ArcheoBERTje predictions and explicit domain knowledge from the thesaurus did not increase the F1 score. We quantitatively and qualitatively analyse the differences between the vocabulary and output of the BERT models on the full collection and provide some valuable insights in the effect of fine-tuning for specific domains. Our results indicate that for a highly specific text domain such as archaeology, further pre-training on domain-specific data increases the model's quality on NER by a much larger margin than shown for other domains in the literature, and that domain-specific pre-training makes the addition of domain knowledge from a thesaurus unnecessary.

show abstract

Making Sense of the Ways We Make Sense of the Past: The Periodo Project

Cited by 19 publications

References 1 publication

Can BERT Dig It? Named Entity Recognition for Information Retrieval in the Archaeology Domain

Can BERT Dig It? Named Entity Recognition for Information Retrieval in the Archaeology Domain

Sentando las bases para una Arqueología Digital en Argentina. El rol de las infraestructuras digitales para la investigación

Can BERT Dig It? -- Named Entity Recognition for Information Retrieval in the Archaeology Domain

Contact Info

Product

Resources

About