Giovanni Colavizza
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Giovanni Colavizza has a total of 112 co-authors. Here are the top co-authors by number of shared publications:
- Ludo Waltman
Leiden University
6 shared publications - Matteo Romanello
University of Lausanne
6 shared publications - Silvio Peroni
University of Bologna
6 shared publications - Frédéric Kaplan
EPFL, Swiss Federal Institute of Technology in Lausanne
4 shared publications - Kevin W. Boyack
SciTech Strategies, Inc., Albuquerque, New Mexico, United States of America
4 shared publications - Nees Jan van Eck
Leiden University, Centre for Science and Technology Studies
4 shared publications
Latest Publications
- Open Access
- 11 September 2026
- Published by Ubiquity Press, Ltd. in Journal of Open Humanities Data
- Vol. 12 (1)
Repurposing and Adapting Language Resources for a Historical Italian Large Language Model Benchmark
We introduce the first benchmark specifically designed to evaluate the capabilities of large language models on historical Italian. While benchmarks for modern Italian are available, historical varieties of the language remain underexplored in this context. To address this gap, we repurpose and adapt several high-quality digital resources to construct a novel benchmark. The result spans a range of tasks, including word sense disambiguation, syntactic structure prediction, intertextuality detection, named entity recognition, authorship verification, and chronological detection, with source texts ranging from the High Middle Ages to the nineteenth century.We evaluate several state-of-the-art large language models in a zero-shot setting. Our findings show that while these models exhibit promising performance, the benchmark remains far from saturated, suggesting that the capabilities of large language models on historical languages still have considerable margin for improvement. Our work offers a standardized evaluation framework to support the selection of models for annotating digital archives and to guide the development of linguistically inclusive language models.
- 11 September 2026
- Published by Ubiquity Press, Ltd. in Journal of Open Humanities Data
- Vol. 12 (1)
Open AccessRepurposing and Adapting Language Resources for a Historical Italian Large Language Model Benchmark
We introduce the first benchmark specifically designed to evaluate the capabilities of large language models on historical Italian. While benchmarks for modern Italian are available, historical varieties of the language remain underexplored in this context. To address this gap, we repurpose and adapt several high-quality digital resources to construct a novel benchmark. The result spans a range of tasks, including word sense disambiguation, syntactic structure prediction, intertextuality detection, named entity recognition, authorship verification, and chronological detection, with source texts ranging from the High Middle Ages to the nineteenth century.We evaluate several state-of-the-art large language models in a zero-shot setting. Our findings show that while these models exhibit promising performance, the benchmark remains far from saturated, suggesting that the capabilities of large language models on historical languages still have considerable margin for improvement. Our work offers a standardized evaluation framework to support the selection of models for annotating digital archives and to guide the development of linguistically inclusive language models.
- Open Access
- 30 June 2026
- Published by MIT Press in Quantitative Science Studies
- Vol. 7, 1-19
An analysis of the effects of open science indicators on citations in the French Open Science Monitor
This study investigates the correlation of citation impact with various open science indicators (OSI) within the French Open Science Monitor (FOSM), a data set comprising approximately 900,000 publications authored by French authors from 2020 to 2022. By integrating data from OpenAlex and Crossref, we analyze open science indicators such as the presence of a pre-print, data sharing, and software sharing in 576,537 publications in the FOSM data set. Our analysis reveals a positive correlation between these OSI and citation counts. Considering our most complete citation prediction model, we find pre-prints are correlated with a significant positive effect of 19% on citation counts, software sharing of 13.5%, and data sharing of 14.3%. We find large variations in the correlations of OSIs with citations in different research disciplines, and observe that the open access status of publications is correlated with an 8.6% increase in citations in our model. While these results remain observational and are limited to the scope of the analysis, they suggest a consistent correlation between citation advantages and open science indicators. Our results may be valuable to policymakers, funding agencies, researchers, publishers, institutions, and other stakeholders who are interested in understanding the academic impacts, or effects, of open science practices.
- 30 June 2026
- Published by MIT Press in Quantitative Science Studies
- Vol. 7, 1-19
Open AccessAn analysis of the effects of open science indicators on citations in the French Open Science Monitor
This study investigates the correlation of citation impact with various open science indicators (OSI) within the French Open Science Monitor (FOSM), a data set comprising approximately 900,000 publications authored by French authors from 2020 to 2022. By integrating data from OpenAlex and Crossref, we analyze open science indicators such as the presence of a pre-print, data sharing, and software sharing in 576,537 publications in the FOSM data set. Our analysis reveals a positive correlation between these OSI and citation counts. Considering our most complete citation prediction model, we find pre-prints are correlated with a significant positive effect of 19% on citation counts, software sharing of 13.5%, and data sharing of 14.3%. We find large variations in the correlations of OSIs with citations in different research disciplines, and observe that the open access status of publications is correlated with an 8.6% increase in citations in our model. While these results remain observational and are limited to the scope of the analysis, they suggest a consistent correlation between citation advantages and open science indicators. Our results may be valuable to policymakers, funding agencies, researchers, publishers, institutions, and other stakeholders who are interested in understanding the academic impacts, or effects, of open science practices.
- Open Access
- 2 June 2026
- Published by Taylor & Francis in Archives and Records
- p. 1-22
AI ready: how to prepare archives for Artificial Intelligence, improve discoverability, and enhance access
- 2 June 2026
- Published by Taylor & Francis in Archives and Records
- p. 1-22
Open AccessAI ready: how to prepare archives for Artificial Intelligence, improve discoverability, and enhance access
- Open Access
- 5 January 2026
- Published by MIT Press in Quantitative Science Studies
- p. 1-16
Wikipedia citations: Reproducible citation extraction from multilingual Wikipedia
Wikipedia is an essential component of the open science ecosystem, yet it is poorly integrated with academic open science initiatives. Wikipedia Citations is a project that focuses on extracting and releasing comprehensive data sets of citations from Wikipedia. A total of 29.3 million citations were extracted from the English Wikipedia in May 2020. Following this one-off research project, we designed a reproducible pipeline that can process any Wikipedia dump in a cloud-based setting. To demonstrate its usability, we extracted 40.6 million citations in February 2023 and 44.7 million citations in February 2024. Furthermore, we equipped the pipeline with an adapted Wikipedia citation template translation module to process multilingual Wikipedia articles in 15 languages so that they are parsed and mapped into a generic structured citation template. This paper presents our open-source software pipeline for retrieving, classifying, and disambiguating citations on demand from a given Wikipedia dump.
- 5 January 2026
- Published by MIT Press in Quantitative Science Studies
- p. 1-16
Open AccessWikipedia citations: Reproducible citation extraction from multilingual Wikipedia
Wikipedia is an essential component of the open science ecosystem, yet it is poorly integrated with academic open science initiatives. Wikipedia Citations is a project that focuses on extracting and releasing comprehensive data sets of citations from Wikipedia. A total of 29.3 million citations were extracted from the English Wikipedia in May 2020. Following this one-off research project, we designed a reproducible pipeline that can process any Wikipedia dump in a cloud-based setting. To demonstrate its usability, we extracted 40.6 million citations in February 2023 and 44.7 million citations in February 2024. Furthermore, we equipped the pipeline with an adapted Wikipedia citation template translation module to process multilingual Wikipedia articles in 15 languages so that they are parsed and mapped into a generic structured citation template. This paper presents our open-source software pipeline for retrieving, classifying, and disambiguating citations on demand from a given Wikipedia dump.
- Open Access
- 21 November 2025
- Published by The Association for Computers and the Humanities
- Vol. 3, 1197-1209
Benchmarking Multimodal Large Language Models in Zero-shot and Few-shot Scenarios: Preliminary Results on Studying Christian Iconography
This study evaluates the capabilities of multimodal large language models (LLMs) in the task of single-label classification of Christian iconography, focusing on their performance in zero-shot and few-shot settings across curated datasets. The goal was to assess whether general-purpose models, such as GPT-4o and Gemini 2.5, can interpret the Iconography, typically addressed by supervised classifiers, and evaluate their performance. Two research questions guided the analysis: (RQ1) How do multimodal LLMs perform on image classification of Christian saints? And (RQ2), how does performance vary when enriching input with contextual information or few-shot exemplars? We conducted a benchmarking study using three datasets supporting Iconclass natively: ArtDL, ICONCLASS, and Wikidata, filtered to include the top 10 most frequent classes. Models were tested under three conditions: (1) classification using class labels, (2) classification with Iconclass descriptions, and (3) few-shot learning with five exemplars. Results were compared against ResNet50 baselines fine-tuned on the same datasets. The findings show that Gemini-2.5 Pro and GPT-4o outperformed the ResNet50 baselines across the three configurations reaching peaks of 90.45% and 88.20% in ArtDL, respectively. Accuracy dropped significantly on the Wikidata dataset, suggesting model sensitivity to image size and metadata alignment. Enriching prompts with class descriptions generally improved zero-shot performance, while few-shot learning produced lower results, with only occasional and minimal increments in accuracy. We conclude that general-purpose multimodal LLMs are capable of classification in visually complex cultural heritage domains, even without specific fine-tuning. However, their performance is influenced by dataset consistency and the design of the prompts. These results support the application of LLMs as metadata curation tools in digital humanities workflows, suggesting future research on prompt optimization and the expansion of the study to other classification strategies and models.
- 21 November 2025
- Published by The Association for Computers and the Humanities
- Vol. 3, 1197-1209
Open AccessBenchmarking Multimodal Large Language Models in Zero-shot and Few-shot Scenarios: Preliminary Results on Studying Christian Iconography
This study evaluates the capabilities of multimodal large language models (LLMs) in the task of single-label classification of Christian iconography, focusing on their performance in zero-shot and few-shot settings across curated datasets. The goal was to assess whether general-purpose models, such as GPT-4o and Gemini 2.5, can interpret the Iconography, typically addressed by supervised classifiers, and evaluate their performance. Two research questions guided the analysis: (RQ1) How do multimodal LLMs perform on image classification of Christian saints? And (RQ2), how does performance vary when enriching input with contextual information or few-shot exemplars? We conducted a benchmarking study using three datasets supporting Iconclass natively: ArtDL, ICONCLASS, and Wikidata, filtered to include the top 10 most frequent classes. Models were tested under three conditions: (1) classification using class labels, (2) classification with Iconclass descriptions, and (3) few-shot learning with five exemplars. Results were compared against ResNet50 baselines fine-tuned on the same datasets. The findings show that Gemini-2.5 Pro and GPT-4o outperformed the ResNet50 baselines across the three configurations reaching peaks of 90.45% and 88.20% in ArtDL, respectively. Accuracy dropped significantly on the Wikidata dataset, suggesting model sensitivity to image size and metadata alignment. Enriching prompts with class descriptions generally improved zero-shot performance, while few-shot learning produced lower results, with only occasional and minimal increments in accuracy. We conclude that general-purpose multimodal LLMs are capable of classification in visually complex cultural heritage domains, even without specific fine-tuning. However, their performance is influenced by dataset consistency and the design of the prompts. These results support the application of LLMs as metadata curation tools in digital humanities workflows, suggesting future research on prompt optimization and the expansion of the study to other classification strategies and models.
- Open Access
- 19 October 2025
- Published by Springer Nature in International Journal on Digital Libraries
- Vol. 26 (4) , 1-17
Deep Learning Approaches to Author Name Disambiguation: A Survey
Purpose: Author Name Disambiguation (AND) is a critical task for digital libraries aiming to link existing authors with their respective publications. Due to the lack of persistent identifiers and intrinsic linguistic challenges such as homonymy, deep learning algorithms have become widespread for addressing this issue. This paper provides a systematic review of state-of-the-art AND techniques based on deep learning within the timeframe 2016-2024, filling a gap in existing literature. Methods: We conducted a systematic literature review using Google Scholar with keywords "author name disambiguation" + "deep learning" and temporal filters. The search yielded 52 documents, of which 28 were selected after full-text assessment. We categorized approaches based on learning paradigms: supervised, unsupervised, and hybrid methods. Additionally, we performed bibliometric analysis using OpenCitations data with network visualization via Gephi. Results: We identified 28 relevant studies employing deep learning for AND. The analysis of the comparable studies reveals that hybrid approaches achieve the best performance, with the top-performing method reaching 89.7 F1-score on AMiner datasets. Supervised methods predominantly focus on author assignment tasks, while unsupervised approaches excel at author grouping. Conclusion: Deep learning methods have significantly impacted AND by enabling integration of structured and unstructured data. Hybrid approaches effectively balance supervised and unsupervised learning, demonstrating superior performance. However, heavy reliance on AMiner datasets and lack of standardized evaluation frameworks remain critical challenges for the field’s advancement.
- 19 October 2025
- Published by Springer Nature in International Journal on Digital Libraries
- Vol. 26 (4) , 1-17
Open AccessDeep Learning Approaches to Author Name Disambiguation: A Survey
Purpose: Author Name Disambiguation (AND) is a critical task for digital libraries aiming to link existing authors with their respective publications. Due to the lack of persistent identifiers and intrinsic linguistic challenges such as homonymy, deep learning algorithms have become widespread for addressing this issue. This paper provides a systematic review of state-of-the-art AND techniques based on deep learning within the timeframe 2016-2024, filling a gap in existing literature. Methods: We conducted a systematic literature review using Google Scholar with keywords "author name disambiguation" + "deep learning" and temporal filters. The search yielded 52 documents, of which 28 were selected after full-text assessment. We categorized approaches based on learning paradigms: supervised, unsupervised, and hybrid methods. Additionally, we performed bibliometric analysis using OpenCitations data with network visualization via Gephi. Results: We identified 28 relevant studies employing deep learning for AND. The analysis of the comparable studies reveals that hybrid approaches achieve the best performance, with the top-performing method reaching 89.7 F1-score on AMiner datasets. Supervised methods predominantly focus on author assignment tasks, while unsupervised approaches excel at author grouping. Conclusion: Deep learning methods have significantly impacted AND by enabling integration of structured and unstructured data. Hybrid approaches effectively balance supervised and unsupervised learning, demonstrating superior performance. However, heavy reliance on AMiner datasets and lack of standardized evaluation frameworks remain critical challenges for the field’s advancement.
- Open Access
Benchmarking Vision-Language and Multimodal Large Language Models in Zero-shot and Few-shot Scenarios: A study on Christian Iconography
This study evaluates the capabilities of Multimodal Large Language Models (LLMs) and Vision Language Models (VLMs) in the task of single-label classification of Christian Iconography. The goal was to assess whether general-purpose VLMs (CLIP and SigLIP) and LLMs, such as GPT-4o and Gemini 2.5, can interpret the Iconography, typically addressed by supervised classifiers, and evaluate their performance. Two research questions guided the analysis: (RQ1) How do multimodal LLMs perform on image classification of Christian saints? And (RQ2), how does performance vary when enriching input with contextual information or few-shot exemplars? We conducted a benchmarking study using three datasets supporting Iconclass natively: ArtDL, ICONCLASS, and Wikidata, filtered to include the top 10 most frequent classes. Models were tested under three conditions: (1) classification using class labels, (2) classification with Iconclass descriptions, and (3) few-shot learning with five exemplars. Results were compared against ResNet50 baselines fine-tuned on the same datasets. The findings show that Gemini-2.5 Pro and GPT-4o outperformed the ResNet50 baselines. Accuracy dropped significantly on the Wikidata dataset, where Siglip reached the highest accuracy score, suggesting model sensitivity to image size and metadata alignment. Enriching prompts with class descriptions generally improved zero-shot performance, while few-shot learning produced lower results, with only occasional and minimal increments in accuracy. We conclude that general-purpose multimodal LLMs are capable of classification in visually complex cultural heritage domains. These results support the application of LLMs as metadata curation tools in digital humanities workflows, suggesting future research on prompt optimization and the expansion of the study to other classification strategies and models.
Open AccessBenchmarking Vision-Language and Multimodal Large Language Models in Zero-shot and Few-shot Scenarios: A study on Christian Iconography
This study evaluates the capabilities of Multimodal Large Language Models (LLMs) and Vision Language Models (VLMs) in the task of single-label classification of Christian Iconography. The goal was to assess whether general-purpose VLMs (CLIP and SigLIP) and LLMs, such as GPT-4o and Gemini 2.5, can interpret the Iconography, typically addressed by supervised classifiers, and evaluate their performance. Two research questions guided the analysis: (RQ1) How do multimodal LLMs perform on image classification of Christian saints? And (RQ2), how does performance vary when enriching input with contextual information or few-shot exemplars? We conducted a benchmarking study using three datasets supporting Iconclass natively: ArtDL, ICONCLASS, and Wikidata, filtered to include the top 10 most frequent classes. Models were tested under three conditions: (1) classification using class labels, (2) classification with Iconclass descriptions, and (3) few-shot learning with five exemplars. Results were compared against ResNet50 baselines fine-tuned on the same datasets. The findings show that Gemini-2.5 Pro and GPT-4o outperformed the ResNet50 baselines. Accuracy dropped significantly on the Wikidata dataset, where Siglip reached the highest accuracy score, suggesting model sensitivity to image size and metadata alignment. Enriching prompts with class descriptions generally improved zero-shot performance, while few-shot learning produced lower results, with only occasional and minimal increments in accuracy. We conclude that general-purpose multimodal LLMs are capable of classification in visually complex cultural heritage domains. These results support the application of LLMs as metadata curation tools in digital humanities workflows, suggesting future research on prompt optimization and the expansion of the study to other classification strategies and models.
- Top Cited PapersOpen Access
- 15 August 2025
- Published by Emerald Publishing in Journal of Documentation
- Vol. 81 (7) , 334-354
Benchmarking large language models for handwritten text recognition
Purpose: The aim of this work is to provide an overview of the current capabilities of Multimodal Large Language Models (MLLMs) for Handwritten Text Recognition (HTR), assessing their potential when compared to traditional task-specific, supervised models. Design/methodology/approach: The approach is that of using a set of openly-available benchmarks to compare different LLMs with strong task-specific supervised baselines for the task of HTR. Findings: The results show that LLMs currently show a strong performance on English texts, yet they demonstrate a weaker performance on languages other than English, and do not possess a significant capability for self-correction. Moreover, their comparison with Transkribus’s models highlight the fact that proprietary LLM models are the best performing, in particular on modern handwriting, while for historical documents the overall performance comparison between LLMs and Transkribus is not consistent. Originality/value: The authors are not aware of a similar study relying on open benchmarks.
- 15 August 2025
- Published by Emerald Publishing in Journal of Documentation
- Vol. 81 (7) , 334-354
Top Cited PapersOpen AccessBenchmarking large language models for handwritten text recognition
Purpose: The aim of this work is to provide an overview of the current capabilities of Multimodal Large Language Models (MLLMs) for Handwritten Text Recognition (HTR), assessing their potential when compared to traditional task-specific, supervised models. Design/methodology/approach: The approach is that of using a set of openly-available benchmarks to compare different LLMs with strong task-specific supervised baselines for the task of HTR. Findings: The results show that LLMs currently show a strong performance on English texts, yet they demonstrate a weaker performance on languages other than English, and do not possess a significant capability for self-correction. Moreover, their comparison with Transkribus’s models highlight the fact that proprietary LLM models are the best performing, in particular on modern handwriting, while for historical documents the overall performance comparison between LLMs and Transkribus is not consistent. Originality/value: The authors are not aware of a similar study relying on open benchmarks.
- Open Access
- 30 October 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (10) , e0311493
An analysis of the effects of sharing research data, code, and preprints on citations
Calls to make scientific research more open have gained traction with a range of societal stakeholders. Open Science practices include but are not limited to the early sharing of results via preprints and openly sharing outputs such as data and code to make research more reproducible and extensible. Existing evidence shows that adopting Open Science practices has effects in several domains. In this study, we investigate whether adopting one or more Open Science practices leads to significantly higher citations for an associated publication, which is one form of academic impact. We use a novel dataset known as Open Science Indicators, produced by PLOS and DataSeer, which includes all PLOS publications from 2018 to 2023 as well as a comparison group sampled from the PMC Open Access Subset. In total, we analyze circa 122’000 publications. We calculate publication and author-level citation indicators and use a broad set of control variables to isolate the effect of Open Science Indicators on received citations. We show that Open Science practices are adopted to different degrees across scientific disciplines. We find that the early release of a publication as a preprint correlates with a significant positive citation advantage of about 20.2% (±.7) on average. We also find that sharing data in an online repository correlates with a smaller yet still positive citation advantage of 4.3% (±.8) on average. However, we do not find a significant citation advantage for sharing code. Further research is needed on additional or alternative measures of impact beyond citations. Our results are likely to be of interest to researchers, as well as publishers, research funders, and policymakers.
- 30 October 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (10) , e0311493
Open AccessAn analysis of the effects of sharing research data, code, and preprints on citations
Calls to make scientific research more open have gained traction with a range of societal stakeholders. Open Science practices include but are not limited to the early sharing of results via preprints and openly sharing outputs such as data and code to make research more reproducible and extensible. Existing evidence shows that adopting Open Science practices has effects in several domains. In this study, we investigate whether adopting one or more Open Science practices leads to significantly higher citations for an associated publication, which is one form of academic impact. We use a novel dataset known as Open Science Indicators, produced by PLOS and DataSeer, which includes all PLOS publications from 2018 to 2023 as well as a comparison group sampled from the PMC Open Access Subset. In total, we analyze circa 122’000 publications. We calculate publication and author-level citation indicators and use a broad set of control variables to isolate the effect of Open Science Indicators on received citations. We show that Open Science practices are adopted to different degrees across scientific disciplines. We find that the early release of a publication as a preprint correlates with a significant positive citation advantage of about 20.2% (±.7) on average. We also find that sharing data in an online repository correlates with a smaller yet still positive citation advantage of 4.3% (±.8) on average. However, we do not find a significant citation advantage for sharing code. Further research is needed on additional or alternative measures of impact beyond citations. Our results are likely to be of interest to researchers, as well as publishers, research funders, and policymakers.
- 23 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (2) , 1409-1418
Partial Annotation Learning for Biomedical Entity Recognition
Named Entity Recognition (NER) is a key task to support biomedical research. In Biomedical Named Entity Recognition (BioNER), obtaining high-quality expert annotated data is laborious and expensive, leading to the development of automatic approaches such as distant supervision. However, manually and automatically generated data often suffer from the unlabeled entity problem , whereby many entity annotations are missing, degrading the performance of full annotation NER models. To conquer this issue, we undertake a systematic exploration of the efficacy of partial annotation learning methods for BioNER, which encompasses a comprehensive evaluation conducted across a spectrum of distinct simulated scenarios of missing entity annotations. Furthermore, we propose a TS-PubMedBERT-Partial-CRF partial annotation learning model. We standardize a compilation of 16 BioNER corpora, encompassing a range of five distinct entity types, to establish a gold standard. And we compare against the state-of-the-art partial annotation model EER-PubMedBERT, the widely acknowledged partial annotation model BiLSTM-Partial-CRF model, and the state-of-the-art full annotation learning BioNER model PubMedBERT tagger. Results show that partial annotation learning-based methods can effectively learn from biomedical corpora with missing entity annotations. Our proposed model outperforms alternatives and, specifically, the PubMedBERT tagger by 38% in F1-score under high missing entity rates. Moreover, the recall of entity mentions in our model demonstrates a competitive alignment with the upper threshold observed on the fully annotated dataset. We have published our data, source code and training records at https://github.com/possible1402/partial\_annotation\_learning .
- 23 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (2) , 1409-1418
Partial Annotation Learning for Biomedical Entity Recognition
Named Entity Recognition (NER) is a key task to support biomedical research. In Biomedical Named Entity Recognition (BioNER), obtaining high-quality expert annotated data is laborious and expensive, leading to the development of automatic approaches such as distant supervision. However, manually and automatically generated data often suffer from the unlabeled entity problem , whereby many entity annotations are missing, degrading the performance of full annotation NER models. To conquer this issue, we undertake a systematic exploration of the efficacy of partial annotation learning methods for BioNER, which encompasses a comprehensive evaluation conducted across a spectrum of distinct simulated scenarios of missing entity annotations. Furthermore, we propose a TS-PubMedBERT-Partial-CRF partial annotation learning model. We standardize a compilation of 16 BioNER corpora, encompassing a range of five distinct entity types, to establish a gold standard. And we compare against the state-of-the-art partial annotation model EER-PubMedBERT, the widely acknowledged partial annotation model BiLSTM-Partial-CRF model, and the state-of-the-art full annotation learning BioNER model PubMedBERT tagger. Results show that partial annotation learning-based methods can effectively learn from biomedical corpora with missing entity annotations. Our proposed model outperforms alternatives and, specifically, the PubMedBERT tagger by 38% in F1-score under high missing entity rates. Moreover, the recall of entity mentions in our model demonstrates a competitive alignment with the upper threshold observed on the fully annotated dataset. We have published our data, source code and training records at https://github.com/possible1402/partial\_annotation\_learning .