Data sensitivity detection in chat interactions for privacy protection

In recent years, there has been exponential growth in using virtual spaces, including dialogue systems, that handle personal information. The concept of personal privacy in the literature is discussed and controversial, whereas, in the technological field, it directly influences the degree of reliability perceived in the information system (privacy ‘as trust’). This work aims to protect the right to privacy on personal data (GDPR, 2018) and avoid the loss of sensitive content by exploring sensitive information detection (SID) task. It is grounded on the following research questions: (RQ1) What does sensitive data mean? How to define a personal sensitive information domain? (RQ2) How to create a state-of-the-art model for SID?(RQ3) How to evaluate the model? RQ1 theoretically investigates the concepts of privacy and the ontological state-of-the-art representation of personal information. The Data Privacy Vocabulary (DPV) is the taxonomic resource taken as an authoritative reference for the definition of the knowledge domain. Concerning RQ2, we investigate two approaches to classify sensitive data: the first - bottom-up - explores automatic learning methods based on transformer networks, the second - top-down - proposes logical-symbolic methods with the construction of privaframe, a knowledge graph of compositional frames representing personal data categories. Both approaches are tested. For the evaluation - RQ3 – we create SPeDaC, a sentence-level labeled resource. This can be used as a benchmark or training in the SID task, filling the gap of a shared resource in this field. If the approach based on artificial neural networks confirms the validity of the direction adopted in the most recent studies on SID, the logical-symbolic approach emerges as the preferred way for the classification of fine-grained personal data categories, thanks to the semantic-grounded tailor modeling it allows. At the same time, the results highlight the strong potential of hybrid architectures in solving automatic tasks.

Fault detection in rotating machines by vibration signal processing techniques

Machines with moving parts give rise to vibrations and consequently noise. The setting up and the status of each machine yield to a peculiar vibration signature. Therefore, a change in the vibration signature, due to a change in the machine state, can be used to detect incipient defects before they become critical. This is the goal of condition monitoring, in which the informations obtained from a machine signature are used in order to detect faults at an early stage. There are a large number of signal processing techniques that can be used in order to extract interesting information from a measured vibration signal. This study seeks to detect rotating machine defects using a range of techniques including synchronous time averaging, Hilbert transform-based demodulation, continuous wavelet transform, Wigner-Ville distribution and spectral correlation density function. The detection and the diagnostic capability of these techniques are discussed and compared on the basis of experimental results concerning gear tooth faults, i.e. fatigue crack at the tooth root and tooth spalls of different sizes, as well as assembly faults in diesel engine. Moreover, the sensitivity to fault severity is assessed by the application of these signal processing techniques to gear tooth faults of different sizes.

Electrochemical sensing strategies for the detection of interactions between biological macromolecules

Electrochemical biosensors provide an attractive means to analyze the content of a biological sample due to the direct conversion of a biological event to an electronic signal, enabling the development of cheap, small, portable and simple devices, that allow multiplex and real-time detection. At the same time nanobiotechnology is drastically revolutionizing the biosensors development and different transduction strategies exploit concepts developed in these field to simplify the analysis operations for operators and end users, offering higher specificity, higher sensitivity, higher operational stability, integrated sample treatments and shorter analysis time. The aim of this PhD work has been the application of nanobiotechnological strategies to electrochemical biosensors for the detection of biological macromolecules. Specifically, one project was focused on the application of a DNA nanotechnology called hybridization chain reaction (HCR), to amplify the hybridization signal in an electrochemical DNA biosensor. Another project on which the research activity was focused concerns the development of an electrochemical biosensor based on a biological model membrane anchored to a solid surface (tBLM), for the recognition of interactions between the lipid membrane and different types of target molecules.

Study of Venturia inaequalis sensitivity to fungicides through molecular and biological methodologies

This study was aimed to correlate the results of relative germination from in vitro tests by trifloxystrobin with those of qPCR on a wide range of V. inaequalis populations and monoconidial isolates. Samples were collected in Italian and Turkish distinct locations from orchards with different scab management. In this study, an allele-specific qPCR with primer sets designed was successfully developed to quantitatively determine the frequency of QoI-resistant allele G143A in populations and monoconidial isolates of V. inaequalis. qPCR followed a similar pattern to that obtained using in vitro conidial germination test in very sensitive and very resistant populations. However, the variability between two test results was observed in hetereogenous populations. Therefore, the results of correlations between in vitro and qPCR showed a positive but not very high correlation for Venturia inaequalis populations (R2=0.70). On the contrary, this correlation between two assessment methods was very high for monoconidial isolates (R2=0.92). Results obtained in quantitative PCR and from traditional spore germination assay differed for the same fungal population and in some cases, it is difficult to assess the resistance in the field by only qPCR. It was concluded that it is not always possible to correlate the frequency of detection of the mutation with biological assessment. In such situations, monitoring by molecular techniques must be supported by standard in vitro resistance assessments and observation of field performance in order to have correct conclusions.

Skin as a promising biomatrix for the early detection of misfolded proteins by RT-QuIC in neurodegenerative diseases

Real-Time Quaking-Induced Conversion (RT-QuIC) is an ultrasensitive assay capable of detecting pathological aggregates of misfolded proteins in biospecimens. In recent years, efforts have been made to find a more feasible and convenient biomatrix as an alternative to CSF, and skin biopsy may be a suitable candidate. This project aimed to evaluate the diagnostic performance of skin RT-QuIC in 3 different cohorts of patients: 1. Creutzfeldt-Jakob disease (CJD), 2. Lewy body disease (LBD), and 3. Isolated REM sleep behavior disorder (iRBD). We studied 71 punch skin samples of 35 patients with CJD, including five assessed in vitam, using 2 two different substrates: Bank vole 23-230 (Bv23-230) and Syrian hamster 23-231 (Ha23-231) recombinant prion protein. Skin prion RT-QuIC showed a 100% specificity with both substrates and a higher sensitivity with the Bv23-230 than Ha23-231 (87.5% vs. 65.6%, respectively). Forty-one patients underwent both lumbar puncture (LB) and skin biopsy; CSF and skin RT-QuIC showed a high level of concordance (38/41, 92.7%). Then, we analyzed samples taken in vitam (n=69) or postmortem (n=49) from patients with Parkinson’s disease (PD), dementia with Lewy bodies (DLB), incidental Lewy body pathology, and neurological controls. Skin α-syn RT-QuIC distinguished LBD patients with an overall accuracy of 94.1% in the two cohorts (sensitivity, 89.2%; specificity, 96.3%). Seventy-nine patients underwent both CSF and skin α-syn RT-QuIC, and the two assays yielded similar diagnostic accuracy (skin, 97.5%; CSF, 98.7%). Finally, we studied 91 iRBD patients and 41 control. In the skin, RT-QuIC showed a sensitivity of 76.9%, specificity of 97.6%, and 82.0% accuracy. 128 participants (88 patients plus 40 controls) underwent both CSF and skin RT-QuIC. The two protocols showed 99.2% of concordance. These works confirmed that skin punch biopsies might represent a valid and convenient alternative to CSF analysis for an early diagnosis of prion diseases and LB-related pathologies.