The Amnesia has been installed in TSD and it is now part of the software portfolio for all the TSD users. The service is documented in the TSD user guide here: https://www.uio.no/english/services/it/research/sensitive-data/use-tsd/software/#toc6 The service can be is used out of the box by any users having structured data, i.e. whenever k-anonymisation is enough. This is the case for all the users that collect data throughout the web-based collection tool built in TSD, namely Nettskjema (). More info about Netskjema Nettskjema here: https://www.uio.no/english/services/it/research/sensitive-data/use-tsd/datacollection/nettskjema/index.html The target now is to test the service with poorly structured data from genomic use case. This is a domain where the k-anynimization algorithm might have deficiency and a more elaborated algorithm might be required. Discussion with the use case provider is on going and the testing will start once the legal aspects has been clarified. Possible design and implementation of new argorithm will be the next phase of the activityAfter importing data from Nettskjema, as JSON or Excel sheets, Amnesia can be used to anonymise the data prior to processing. Usage: One use case for Amnesia in TSD is for anonymising personal data where there is a lack of data subject consent for fully viewing and processing personal data. Another use case has been for moving the sensitive data out of the secure e-Infrastructure as "results" for research publication and/all archiving. Lacking features: In some use cases, anonymization of audio and video material is necessary, which is not supported by Amnesia. The usage of Amnesia for unstructured genomic data, which would be useful for many use cases, is not applicable. Amnesia can be used for structured data anonymization only Conclusion: Amnesia is a useful user friendly tool for anonymising structured data. This functionality is useful for many sensitive data use cases. Amnesia however is not supporting anonymization of unstructured genomic data and video/audio data. Possible design and implementation of new algorithms can be useful for supporting different types of data. |