Giovanni Colavizza

Giovanni Colavizza, Lauren Cadwallader, Iain Hrynaszkiewicz

This study investigates the correlation of citation impact with various open science indicators (OSI) within the French Open Science Monitor (FOSM), a dataset 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 dataset. 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 open access status of publications is correlated with a 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 policy makers, funding agencies, researchers, publishers, institutions, and other stakeholders who are interested in understanding the academic impacts, or effects, of open science practices.

Giorgia Crosilla, Lukas Klic, Giovanni Colavizza

Traditional machine learning models for Handwritten Text Recognition (HTR) rely on supervised training, requiring extensive manual annotations, and often produce errors due to the separation between layout and text processing. In contrast, Multimodal Large Language Models (MLLMs) offer a general approach to recognizing diverse handwriting styles without the need for model-specific training. The study benchmarks various proprietary and open-source LLMs against Transkribus models, evaluating their performance on both modern and historical datasets written in English, French, German, and Italian. In addition, emphasis is placed on testing the models' ability to autonomously correct previously generated outputs. Findings indicate that proprietary models, especially Claude 3.5 Sonnet, outperform open-source alternatives in zero-shot settings. MLLMs achieve excellent results in recognizing modern handwriting and exhibit a preference for the English language due to their pre-training dataset composition. Comparisons with Transkribus show no consistent advantage for either approach. Moreover, LLMs demonstrate limited ability to autonomously correct errors in zero-shot transcriptions.

Yurui Zhu, Giovanni Colavizza, Matteo Romanello

Bibliographic reference extraction and parsing are foundational for citation indexing, linking, and downstream scholarly knowledge-graph construction. However, most established evaluations focus on clean, English, end-of-document bibliographies, and therefore underrepresent the Social Sciences and Humanities (SSH), where citations are frequently multilingual, embedded in footnotes, abbreviated, and shaped by heterogeneous historical conventions. We present a unified benchmark that targets these SSH-realistic conditions across three complementary datasets: CEX (English journal articles spanning multiple disciplines), EXCITE (German/English documents with end-section, footnote-only, and mixed regimes), and LinkedBooks (humanities references with strong stylistic variation and multilinguality). We evaluate three tasks of increasing difficulty -- reference extraction, reference parsing, and end-to-end document parsing -- under a schema-constrained setup that enables direct comparison between a strong supervised pipeline baseline (GROBID) and contemporary LLMs (DeepSeek-V3.1, Mistral-Small-3.2-24B, Gemma-3-27B-it, and Qwen3-VL (4B-32B variants)). Across datasets, extraction largely saturates beyond a moderate capability threshold, while parsing and end-to-end parsing remain the primary bottlenecks due to structured-output brittleness under noisy layouts. We further show that lightweight LoRA adaptation yields consistent gains -- especially on SSH-heavy benchmarks -- and that segmentation/pipelining can substantially improve robustness. Finally, we argue for hybrid deployment via routing: leveraging GROBID for well-structured, in-distribution PDFs while escalating multilingual and footnote-heavy documents to task-adapted LLMs.

Francesca Cappelli, Giovanni Colavizza, Silvio Peroni

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 used by researchers and the presence of intrinsic linguistic challenges, such as homonymy, the development of Deep Learning algorithms to address this issue has become widespread. Many AND deep learning methods have been developed, and surveys exist comparing the approaches in terms of techniques, complexity, performance. However, none explicitly addresses AND methods in the context of deep learning in the latest years (i.e. timeframe 2016-2024). In this paper, we provide a systematic review of state-of-the-art AND techniques based on deep learning, highlighting recent improvements, challenges, and open issues in the field. We find that DL methods have significantly impacted AND by enabling the integration of structured and unstructured data, and hybrid approaches effectively balance supervised and unsupervised learning.

Gianmarco Spinaci, Lukas Klic, Giovanni Colavizza

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.

Shibingfeng Zhang, Giovanni Colavizza

Large language models have demonstrated remarkable versatility across a wide range of natural language processing tasks and domains. One such task is Named Entity Recognition (NER), which involves identifying and classifying proper names in text, such as people, organizations, locations, dates, and other specific entities. NER plays a crucial role in extracting information from unstructured textual data, enabling downstream applications such as information retrieval from unstructured text. Traditionally, NER is addressed using supervised machine learning approaches, which require large amounts of annotated training data. However, historical texts present a unique challenge, as the annotated datasets are often scarce or nonexistent, due to the high cost and expertise required for manual labeling. In addition, the variability and noise inherent in historical language, such as inconsistent spelling and archaic vocabulary, further complicate the development of reliable NER systems for these sources. In this study, we explore the feasibility of applying LLMs to NER in historical documents using zero-shot and few-shot prompting strategies, which require little to no task-specific training data. Our experiments, conducted on the HIPE-2022 (Identifying Historical People, Places and other Entities) dataset, show that LLMs can achieve reasonably strong performance on NER tasks in this setting. While their performance falls short of fully supervised models trained on domain-specific annotations, the results are nevertheless promising. These findings suggest that LLMs offer a viable and efficient alternative for information extraction in low-resource or historically significant corpora, where traditional supervised methods are infeasible.

Liangping Ding, Giovanni Colavizza, Zhixiong Zhang

Motivation: 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. Results: To address this problem, we systematically study the effectiveness of partial annotation learning methods for biomedical entity recognition over different simulated scenarios of missing entity annotations. Furthermore, we propose a TS-PubMedBERT-Partial-CRF partial annotation learning model. We harmonize 15 biomedical NER corpora encompassing five entity types to serve as a gold standard and compare against two commonly used partial annotation learning models, BiLSTM-Partial-CRF and EER-PubMedBERT, 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. The recall of entity mentions in our model is also competitive with the upper bound on the fully annotated dataset.

Konstantin Todorov, Giovanni Colavizza

Neural language models are the backbone of modern-day natural language processing applications. Their use on textual heritage collections which have undergone Optical Character Recognition (OCR) is therefore also increasing. Nevertheless, our understanding of the impact OCR noise could have on language models is still limited. We perform an assessment of the impact OCR noise has on a variety of language models, using data in Dutch, English, French and German. We find that OCR noise poses a significant obstacle to language modelling, with language models increasingly diverging from their noiseless targets as OCR quality lowers. In the presence of small corpora, simpler models including PPMI and Word2Vec consistently outperform transformer-based models in this respect.

Lucas G. S. Jeub, Giovanni Colavizza, Xiaowen Dong et al.

We propose a decentralised "local2global"' approach to graph representation learning, that one can a-priori use to scale any embedding technique. Our local2global approach proceeds by first dividing the input graph into overlapping subgraphs (or "patches") and training local representations for each patch independently. In a second step, we combine the local representations into a globally consistent representation by estimating the set of rigid motions that best align the local representations using information from the patch overlaps, via group synchronization. A key distinguishing feature of local2global relative to existing work is that patches are trained independently without the need for the often costly parameter synchronization during distributed training. This allows local2global to scale to large-scale industrial applications, where the input graph may not even fit into memory and may be stored in a distributed manner. We apply local2global on data sets of different sizes and show that our approach achieves a good trade-off between scale and accuracy on edge reconstruction and semi-supervised classification. We also consider the downstream task of anomaly detection and show how one can use local2global to highlight anomalies in cybersecurity networks.

Lucas G. S. Jeub, Giovanni Colavizza, Xiaowen Dong et al.

We propose a decentralised "local2global" approach to graph representation learning, that one can a-priori use to scale any embedding technique. Our local2global approach proceeds by first dividing the input graph into overlapping subgraphs (or "patches") and training local representations for each patch independently. In a second step, we combine the local representations into a globally consistent representation by estimating the set of rigid motions that best align the local representations using information from the patch overlaps, via group synchronization. A key distinguishing feature of local2global relative to existing work is that patches are trained independently without the need for the often costly parameter synchronisation during distributed training. This allows local2global to scale to large-scale industrial applications, where the input graph may not even fit into memory and may be stored in a distributed manner. Preliminary results on medium-scale data sets (up to $\sim$7K nodes and $\sim$200K edges) are promising, with a graph reconstruction performance for local2global that is comparable to that of globally trained embeddings. A thorough evaluation of local2global on large scale data and applications to downstream tasks, such as node classification and link prediction, constitutes ongoing work.

Kasra Hosseini, Kaspar Beelen, Giovanni Colavizza et al.

We present four types of neural language models trained on a large historical dataset of books in English, published between 1760-1900 and comprised of ~5.1 billion tokens. The language model architectures include static (word2vec and fastText) and contextualized models (BERT and Flair). For each architecture, we trained a model instance using the whole dataset. Additionally, we trained separate instances on text published before 1850 for the two static models, and four instances considering different time slices for BERT. Our models have already been used in various downstream tasks where they consistently improved performance. In this paper, we describe how the models have been created and outline their reuse potential.