Abstract:
The majority of the cultural objects in museums stay in storage areas, placed inside different types of open and closed containers. A common problem, especially for small and middle-size museums, is to understand if objects are well conserved, or if they are affected by some agents of deterioration as, for instance, incorrect temperature or relative humidity. This thesis presents the framework and implementation of the APACHE Decision Support System (DSS) developed under the "Active & intelligent PAckaging materials and display cases as a tool for preventive conservation of Cultural HEritage" (APACHE) project, funded by the Horizon 2020 European program. The APACHE DSS is developed to help cultural institutions to understand the possible threats that may affect a collection and guide them to the selection of the best preventive measure to apply. To achieve this goal, the APACHE DSS records environmental data that come as time series from the sensors installed in object's containers. This data is processed, and if some threats are identified, the APACHE DSS suggests a ranked list of preventive measures that address that threats.
The use of compression techniques can improve the efficiency of sensor data management, streaming, and storage. For this reason, the second part of the thesis focuses on methods for the online compression of time series. In addition to analysing the relevant literature in this field, it also presents a novel lossy compression method based on the use of neural networks to learn how to compress time series in a lossy way while preserving the accuracy of classification based on compressed data.
