National Technical University of Athens (NTUA)
Πρότυπο πληροφοριακό σύστημα διαχείρισης, επεξεργασίας και ανάλυσης μεγάλων πολυδιάστατων δεδομένων για τη διάγνωση των ιστορικών μνημείων
Abstract
dc:descriptionThis doctoral dissertation addresses the design, experimental development, and evaluation of an Integrated Information System for the management, processing, semantic integration, spatial correlation, analysis, and visualization of large multidimensional datasets in the diagnosis of historic monuments. The research starts from a critical challenge of contemporary scientific documentation in conservation projects: data are increasingly multimodal, heterogeneous, spatially complex, and methodologically fragmented. They are produced by different scientific teams, project phases, measurement techniques, and interpretive levels. As a result, simple file archiving is insufficient. What is required is an integrated information environment capable of preserving the relationship between each record and its project phase, monument area, measurement method, provenance, processing parameters, and scientific meaning. The restoration project of the Holy Aedicule of the Holy Sepulchre in Jerusalem, carried out in 2016-2017 under the coordination of the National Technical University of Athens, is used as the principal case study. The monument brings together exceptional historical, religious, architectural, and technical complexity, while its material fabric incorporates successive phases from late antiquity to the modern period. The project generated an extensive and heterogeneous body of evidence, including three-dimensional geometric documentation, photogrammetric models, high-resolution imagery, ground-penetrating radar and other survey data, infrared thermography, ultrasonic and environmental measurements, materials analyses, historical records, drawings, reports, field observations, and derived processing results. Bringing these data into a common framework of management and interpretation is essential for understanding conservation state, deterioration mechanisms, and the relationships among material fabric, pathology, and interventions. The proposed approach goes beyond the model of a digital repository and treats the information system as a research infrastructure for data governance. Its architecture is organized around the triple correlation of project phase, monument area, and measurement technique, so that each dataset can be situated within a specific temporal, spatial, and methodological context. The data layer combines structured metadata management, geospatial information, large digital-object storage, and a graph database for representing complex relationships among monument parts, structural elements, samples, measurements, observations, documents, and analytical results. At the same time, full-text retrieval, access control, traceability, versioning, action history, and provenance documentation mechanisms are integrated so that the system supports not only storage, but also reliability, reuse, and controlled scientific exploitation of information. A central contribution of the dissertation is the semantic integration of the data through an ontological model and knowledge graph aligned with international standards for cultural heritage documentation and scientific observation. The conceptual model represents relationships among project phases, monument areas, structural elements, samples, materials, measurement techniques, datasets, digital documents, observations, analytical processes, and generated results. In this way, information does not remain isolated at the level of individual files, but is connected to its physical, spatial, temporal, and scientific meaning. The approach enables queries and correlation paths that cross different documentation levels, such as moving from a sample or monument area to the related measurements, images, material analyses, survey data, observations, and derived results. Particular emphasis is placed on the spatial correlation of information with the three-dimensional model of the monument. The dissertation develops mechanisms for three-dimensional visualization, point selection, polygonal regions of interest, zones, and cross-sections, linking data to actual positions on the monument geometry. Regions of interest are not treated merely as two-dimensional projections or as user-defined control vertices. They are transformed into spatial representations that follow the real surface of the three-dimensional model through sampling of surface geometry and scene depth. As a result, annotation, search, and data projection can be associated with stone blocks, panels, facades, pathology zones, or measurement areas while preserving their connection to the three-dimensional material reality of the monument. This spatial infrastructure is used to correlate geometric documentation with non-destructive testing and imaging techniques. The dissertation examines the association of survey measurements, such as ground-penetrating radar, with specific monument areas and depth levels, as well as the projection and interpretation of infrared thermographic data in relation to the three-dimensional geometry. This enhances the interpretation of phenomena such as moisture, thermal discontinuities, internal variations, material discontinuities, and potential pathology zones. Correlating different data types on the same spatial substrate enables comparison of observations produced by different techniques and strengthens evidence-based decision-making in contexts where physical access to the monument is limited and conservation interventions carry a high degree of responsibility. The system also supports data processing and analysis workflows, with emphasis on transparent documentation of steps, parameters, and generated results. The experimental implementation includes procedures for cleaning, transformation, feature selection, normalization, training, and evaluation of machine-learning models, as well as applications to thermographic data, survey data, and multidimensional measurements. The value of the analytical mechanisms does not lie only in producing predictions or classifications. It lies mainly in connecting analytical results back to the primary record, spatial reference, project phase, processing parameters, and interpretive framework. In this way, analysis becomes a documented scientific process within the information system rather than an isolated computational task. The dissertation contributes theoretically, methodologically, and practically. At the theoretical level, it proposes a coherent framework for governance, semantic organization, and spatial referencing of multimodal cultural heritage data. At the methodological level, it documents a process for transforming the real needs of an interdisciplinary restoration project into requirements, architecture, data models, workflows, and evaluation mechanisms. At the practical level, it presents a functional research prototype that demonstrates integrated large-scale data management, spatial and semantic correlation, three-dimensional visualization, analytical processing, and documented retrieval. Its application to the Holy Aedicule shows that the integration of data, geometry, semantics, and analytical workflows can substantially support the diagnosis, restoration, and long-term preservation of complex historic monuments.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Alexakis, Emmanouil
- Αλεξάκης, Εμμανουήλ
Subjects
dc:subject × 36- Πληροφοριακά συστήματα
- Διακυβέρνηση δεδομένων
- Πολυτροπικά δεδομένα
- Σημασιολογική διαλειτουργικότητα
- Μηχανική μάθηση
- Βαθειά μάθηση
- Τρισδιάστατη οπτικοποίηση
- Πολιτιστική κληρονομιά
- Ιερό Κουβούκλιο Παναγίου Τάφου
- Μη καταστρεπτικοί έλεγχοι
- Θερμογραφία υπερύθρου
- Γεωραντάρ
- Information systems
- Data Governance
- Multimodal data
- Semantic interoperability
- Machine learning
- Deep learning
- Three-dimensional visualization
- Cultural heritage
- Holy Aedicule of the Holy Sepulchre
- Non-Destructive Testing (NDT)
- Infrared thermography
- Ground-penetrating radar
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Φυσικές Επιστήμες
- Τεχνητή νοημοσύνη
- Μηχανική Υλικών
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη υλικών, διεπιστημονική προσέγγιση
- Computer and Information Sciences
- Natural Sciences
- Artificial Intelligence
- Materials Engineering
- Engineering and Technology
- Materials science, interdisciplinary
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/62782