Document Intelligence Centre of Excellence

National Center for Scientific Research Demokritos
Patr. Gregoriou E & Neapoleos Str 27
15341 Agia Paraskevi
Athens, Greece

DICE is a Document Intelligence Centre of Excellence. Its mission is to develop novel AI methods and become a world class reference centre in AI for business documents.

Its vision is to connect researchers, scientists and AI professionals with business experts from a wide range of industrial sectors, aiming at using emerging technologies to accelerate innovation.

Research synopsis: Our principal research interests lie in the development of machine learning and statistical methodologies for solving problems involving un-/semi-/fully-structured financial data.

We are part of the Institute of Informatics and Telecommunications (IIT) at the National Center for Scientific Research - NCSR “Demokritos”.

Check out our publications and our projects outreach to learn more abour our work and feel free to reach out, if you are interested!

News

Jul 03, 2026	Congratulations to Dimitris Kelesis who awarded Best PhD Dissertation for his PhD Thesis Efficient Approaches to Deal with Oversmoothing in Deep Graph Neural Networks from the Hellenic Artificial Intelligence Society (EETN)🍾!
Jun 03, 2026	FinGuard on Science Agora Proof-of-Concept Day
May 25, 2026	FinGuard was selected to be presented in the Science Agora Proof-of-Concept for Emerging Technologies 👏🏽! Looking forward to connecting and seeing other interesting ideas!
May 17, 2026	Presentation on IIT's Information Day

Selected Publications

ICAIF
A Multimodal Alignment-Based Anomaly Detection Method for Bankruptcy Prediction

Andreas Sideras, Konstantinos Bougiatiotis, Elias Zavitsanos, Georgios Paliouras, and George Vouros

In , Singapore, Nov 2025

Abs DOI Bib PDF Code

We present a novel anomaly detection method for next-year bankruptcy prediction, utilizing a combination of financial figures and textual content from annual reports. Our approach, MABAD, learns a shared representation space where non-bankrupt firms share position and orientation. Samples that deviate from this pattern are assigned a higher anomaly score. The proposed method is tailored for highly imbalanced scenarios and is robust to heterogeneous, incomplete, and potentially contradictory inputs. We demonstrate that MABAD consistently outperforms a range of strong baselines, and we also curate and release a new publicly available multisource dataset to foster further research in the domain.
@inproceedings{10.1145/3768292.3770380, author = {Sideras, Andreas and Bougiatiotis, Konstantinos and Zavitsanos, Elias and Paliouras, Georgios and Vouros, George}, title = {A Multimodal Alignment-Based Anomaly Detection Method for Bankruptcy Prediction}, year = {2025}, isbn = {9798400722202}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, url = {https://aclanthology.org/2022.acl-long.303/}, doi = {10.1145/3768292.3770380}, pages = {53–61}, numpages = {9}, keywords = {Bankruptcy Prediction, Financial Dataset, Anomaly Detection, Multimodal Machine Learning}, location = {Singapore}, series = {ICAIF '25}, month = nov, }
ICAIF
Positive-Unlabeled Learning for Financial Misstatement Detection under Realistic Constraints

Elias Zavitsanos, Konstantinos Bougiatiotis, Andreas Sideras, and Georgios Paliouras

In Proceedings of the 6th ACM International Conference on AI in Finance, Singapore, Nov 2025

Abs DOI Bib PDF Code

Detecting financial misstatements is critical for market integrity but remains challenging due to class imbalance, delayed discovery, and limited labeled data. We propose a novel Positive-Unlabeled (PU) learning framework that models the detection task under realistic constraints, where only a small subset of misstatements is known at training time. Our approach integrates unlabeled data into training, preserves temporal structure, and accounts for extreme imbalance. We construct and release a benchmark dataset reflecting these characteristics and evaluate several PU learning methods against recent baselines. Results show that PU-based models consistently outperform supervised approaches, highlighting their suitability for real-world misstatement detection.
@inproceedings{10.1145/3768292.3770366, author = {Zavitsanos, Elias and Bougiatiotis, Konstantinos and Sideras, Andreas and Paliouras, Georgios}, title = {Positive-Unlabeled Learning for Financial Misstatement Detection under Realistic Constraints}, year = {2025}, isbn = {9798400722202}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, url = {https://aclanthology.org/2022.acl-long.303/}, doi = {10.1145/3768292.3770366}, booktitle = {Proceedings of the 6th ACM International Conference on AI in Finance}, pages = {864–872}, numpages = {9}, keywords = {Misstatement Detection, PU Learning, 10-K, Risk Assessment}, location = {Singapore}, series = {ICAIF '25}, month = nov, }
ACM
Machine Learning for Identifying Risk in Financial Statements: A Survey

Elias Zavitsanos, Eirini Spyropoulou, George Giannakopoulos, and Georgios Paliouras

ACM Comput. Surv., Mar 2025

Abs DOI Bib

The work herein reviews the scientific literature on Machine Learning approaches for financial risk assessment using financial reports. We identify two prominent use cases that constitute fundamental risk factors for a company, namely misstatement detection and financial distress prediction. We further categorize the related work along four dimensions that can help highlight the peculiarities and challenges of the domain. Specifically, we group the related work based on (a) the input features used by each method, (b) the sources providing the labels of the data, (c) the evaluation approaches used to confirm the validity of the methods, and (d) the machine learning methods themselves. This categorization facilitates a technical overview of risk detection methods, revealing common patterns, methodologies, significant challenges, and opportunities for further research in the field.
@article{10.1145/3723157, title = {Machine Learning for Identifying Risk in Financial Statements: A Survey}, author = {Zavitsanos, Elias and Spyropoulou, Eirini and Giannakopoulos, George and Paliouras, Georgios}, year = {2025}, month = mar, journal = {ACM Comput. Surv.}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, doi = {10.1145/3723157}, issn = {0360-0300}, url = {https://doi.org/10.1145/3723157}, keywords = {risk assessment, misstatement detection, financial distress, bankruptcy prediction, fraud detection, financial reports, financial statements, machine learning, data mining, auditing}, }

ICAIF

Bankruptcy Prediction: Data Augmentation, LLMs and the Need for Auditor’s Opinion

Andreas Sideras, Konstantinos Bougiatiotis, Elias Zavitsanos, Georgios Paliouras, and George Vouros

In Proceedings of the 5th ACM International Conference on AI in Finance, Nov 2024

Bib PDF Code Website

@inproceedings{sideras2024bankruptcy,
  title = {Bankruptcy Prediction: Data Augmentation, LLMs and the Need for Auditor's Opinion},
  author = {Sideras, Andreas and Bougiatiotis, Konstantinos and Zavitsanos, Elias and Paliouras, Georgios and Vouros, George},
  year = {2024},
  booktitle = {Proceedings of the 5th ACM International Conference on AI in Finance},
  pages = {453--460},
  month = nov,
}

Nature

Entrant: A large financial dataset for table understanding

Elias Zavitsanos, Dimitris Mavroeidis, Eirini Spyropoulou, Manos Fergadiotis, and Georgios Paliouras

Scientific Data, Aug 2024

Bib PDF Code

@article{zavitsanos2024entrant,
  title = {Entrant: A large financial dataset for table understanding},
  author = {Zavitsanos, Elias and Mavroeidis, Dimitris and Spyropoulou, Eirini and Fergadiotis, Manos and Paliouras, Georgios},
  year = {2024},
  month = aug,
  journal = {Scientific Data},
  publisher = {Nature Publishing Group UK London},
  volume = {11},
  number = {1},
  pages = {876},
}