MAPPALAB at CAA 2026
From March 31 to April 4, MAPPALAB will take part in CAA 2026, the 53rd edition of the conference on Computer Applications and Quantitative Methods in Archaeology.
The poster session will be held on Wednesday, April 1 from 10:30 to 14:30. The posters will remain on display from Wednesday, April 1 to Friday, April 3. The posters we will present are:
– Elisa Paperini: “Generative Artificial Intelligence to simulate ancient environmental landscapes”. The poster explores the use of predictive and generative AI for the visual reconstruction of ancient plant landscapes from European pollen data. This approach offers new potential for research and outreach, transforming datasets into realistic outputs. At the same time, it highlights the need to balance the strong realism of AI-generated images with the scientific uncertainty inherent in palaeoenvironmental reconstructions.
– Federica Mauro, Franco Cicirelli, Ettore Ritacco, Gabriele Gattiglia: “A Fine-Grained Pottery Dataset Through the Lens of Siamese Networks: Preliminary Results”. Fine-grained pottery classification with a ResNet101 reaches approximately 74% accuracy over 86 classes, but struggles with rare and visually similar classes. A Siamese network, though less effective on its own, captures complementary similarity patterns and can correct uncertain predictions, suggesting its value within a more integrated framework.
– Lorena Bravi, Martina Naso, Massimiliano Puntin: “Linking Analytical Data through Semantics: Challenges and Perspectives in Archaeometric Research”. The contribution proposes a methodological framework based on ontologies and graph models to represent in a structured and interoperable way the various analytical phases (acquisition, calibration, quantification and interpretation), with the aim of improving transparency, traceability and data reuse. The participation is part of their doctoral research activities and responds to the need to address the fragmentation and heterogeneity of archaeometric data, promoting the adoption of shared standards and advanced digital tools. It also contributes to the international debate on digital infrastructures for cultural heritage, strengthening the integration between laboratory analysis and archaeological interpretation.
The talks we will give are:
– Elisa Paperini: “A Survey on Attitudes towards Artificial Intelligence in Archaeology.” This contribution presents the process of creating a collaborative survey, developed within the COST Action MAIA (CA23141), which investigates how Artificial Intelligence is perceived, adopted and understood across the various fields of archaeology. The study promotes interdisciplinary dialogue to develop shared standards for a responsible and informed use of AI in archaeological research.
-Martina Naso, Nevio Dubbini: “Beyond the Instruments: AI-Driven Integration of Multisensor Data in Archaeometry”. The workflow developed within the AUTOMATA project for the integration of multisensor archaeometric data obtained through non-destructive techniques (HSI, pXRF and Raman). HSI is used as a first screening layer to analyse the entire surface of the fragments and guide the selection of the most significant areas to be investigated with point-based techniques. The contribution also shows how certain stages of the process, such as the selection of regions of interest and data quality assessment, can be supported by AI-based approaches, facilitating the integration and interpretation of data from different techniques.
-Nevio Dubbini, Francesco D’Antoni: “Extracting archaeological knowledge from legacy records: a human-in-the-loop approach using AI and NLP”. A hybrid AI–human workflow for extracting structured archaeological data from unstructured archival records, combining OCR, NLP and large language models with expert validation.
-Nevio Dubbini: “Automated Segmentation and Integration of Avifaunal Bone Image Datasets Using Deep Learning-Based Mask Generation”. This presentation concerns a deep learning–based workflow for the automatic segmentation and standardisation of avian bone images from heterogeneous datasets.
