Archivio istituzionale della ricerca - Alma Mater Studiorum Università di Bologna
Archivio istituzionale della ricerca - Alma Mater Studiorum Università di BolognaNot a member yet
336958 research outputs found
Sort by
Detecting Smart Contract Vulnerabilities using Transformers and LLMs
No full textThis study investigates the detection of vulnerabilities in smart contracts using various transformer models and Large Language Model (LLM) systems. We evaluated BERT, CodeBERT, DistilBERT, and the Gemini model, employing techniques such as aggregation of chunks to enhance performance. The results indicate that simple transformers applied to source code generally perform worse than when applied to byte-code. However, the use of aggregation techniques on the source code significantly improved the model performance. We also evaluate the use of meta-classifiers for multimodal data, by stacking multiple transformers working on source code and byte-code. The Random Forest meta-classifier achieved the highest performance but exhibited significant overfitting. The Gemini model demonstrates limited performance, highlighting the necessity of proper training for LLM systems
Nota del curatore
No full textIl testo presenta il contenuto del volume dedicato a “Musica e liturgia in Italia nel Sei e Settecento”. Gli studiosi che hanno aderito al progetto si sono confrontati su alcuni importanti temi legati alla musica in rapporto alla liturgia in Italia nei secoli XVII e XVIII, lungo alcune direttrici principali: quali erano gli stili musicali praticati e le tradizioni cui essi rispondevano? in che misura essi risentivano delle consuetudini locali? qual era il ruolo dell’apparato cerimoniale nella definizione e nella fisionomia delle musiche da impiegare nelle celebrazioni? in che modo gli spazi cui le musiche erano destinate giocavano un loro ruolo
Nonessential pollutant metals in the food chain: state of art and particular focus on thallium and platinum group metals (PGMs)
No full textUnfortunately, the legislation currently in force imposes maximum concentration limits for these metals only for very few elements and only for very few types of food matrices. This chapter therefore wants to bring attention to the great problem of toxic metals, which are little taken into consideration by the regulations of International Organizations for the protection of human health. The same regulations, and rightly so, consider in an extremely detailed manner the numerous organic compounds that can be present as truly dangerous pollutants in food matrices placed on the market for sale and human consumption. It should therefore be underlined that greater attention by International Organizations to the inorganic part, specifically toxic metals, especially nonessential ones, would be not only very important, but also decidedly desirable. Therefore, the relevant International Organizations should have among their main objectives precisely that of filling this gap, i.e., the lack of clear and, above all, exhaustive laws regarding the maximum allowable concentrations in matrices involved in the food chain, expanding to the maximum both the number of metals taken into consideration and the types of metrics considered. This work aims to highlight the state of the art regarding the presence of nonessential, and therefore toxic and potentially very dangerous, metals in the food chain. The metals taken into consideration in this discussion are only marginally the nonessential and toxic metals par excellence—Pb, Cd and Hg—of which there is a vast literature, but some future food metal contaminants, absolutely nonessential and which have appeared in the environment only recently, starting from the end of the past century, i.e., thallium and platinum group metals (PGMs): Pt, Pd, Rh, Os, Ru, and Ir. The types of matrices considered are: sea food (mussels, clams, oysters, fishes and algae), wine, vegetables, meals, meat, milk and milk derivatives, tea, and herbal medicines
Midwives’ perspectives on exploring emotional well-being during prenatal and postnatal care in the Italian contest: a qualitative study (the BEN_EM_OS study)
Get PDFProblem: Perinatal mental health issues, including depression, anxiety disorders, and post-traumatic stress disorder, affect 10-20% of women during pregnancy and postpartum. Despite routine healthcare contact, 50% of
cases go undetected and untreated.
Background: Maternal health issues are crucial for maternal well-being, infant development, and family functioning. However, gaps exist in screening practices and service provision across healthcare systems. In Italy,
midwives provide comprehensive care, emotional support, and risk factor evaluation. The Italian Ministry of
Health launched a project in 2016 to assess psychological well-being during antenatal and postnatal care, and
midwives play a crucial role in screening women at risk of mental health difficulties.
Aim: This study aims to explore Italian midwives’ perspectives on barriers and facilitators to mental health
screening during prenatal and postnatal care.
Methods: This study uses Thorne’s interpretive description methodology. A sample of 24 midwives was recruited.
Information about their experiences and strategies used during the mental health wellbeing screening was
explored through in-depth interviews. The constant comparative method was used to analyze the data
Findings: According to the study funding, midwives address the topic of emotional well-being throughout
pregnancy and the postnatal period by drawing on five professional knowledge pillars: Maieutic Communication,
Waiting for the right moment, Empathy, Empowerment, Intuition
Conclusion: This study highlights the importance of assessing emotional and psychological well-being during
pregnancy and postpartum, with midwives playing a crucial rol
GENERATIVE AI MEETS ARCHITECTURE: TRANSFORMING VISIONS FROM TEXT, SKETCHES, AND 3D MODELS
No full textGenerative AI is transforming architectural design, enabling everything from text-based concept generation to sketch
animation and hyper-realistic renderings. This study examines whether Gen AI can go beyond visual inspiration to actually
generate viable architectural designs. Twenty-one tools were tested using three input types, text prompts, sketches, and 3D
models, across specific case studies. A total of 25 renderings were produced and analyzed. The evaluation focused on how
well the outputs aligned with the inputs, as well as their structural and functional plausibility. Tools were assessed for their
creative flexibility, fidelity to prompts, and architectural coherence. Findings reveal both the potential and current limitations
of Gen AI in architectural practice
Monitoring forest attributes, C-fluxes, and C-stocks using the process-based model 3D-CMCC-FEM at the National level
Get PDFProcess-based forest models (PBFMs) are valuable tools for investigating the effects of climate change and alternative forest management strategies. However, they can also be considered a tool for monitoring forest conditions over short to extended periods, when ancillary data are scarce and continuous measurements are time-consuming. This study aims to evaluate the PBFM named ‘3D-CMCC-FEM’ on its capacity to monitor Italian forests. We simulated 5135 plots, corresponding to ∼83 % of the 6174 field plots included in the second Italian National Forest Inventory (NFI). The model was used to predict the carbon, nitrogen, and water cycles, including structural variables, and validated against observations from the third NFI. We also compared gross primary productivity (GPP) with two well-known remote sensing-based (RS) datasets. Overall, the model showed good performance in reproducing aboveground stocks and structural variables, with r2 values ranging from 0.65 for diameter to 0.49 for height, and RMSE% ranging from 32 % for diameter and height to 46 % for volume. We aggregated and validated the simulation at the NUT2 level against the estimate of the third NFI, obtaining higher accuracy than the plot-level validation. Compared to RS–data the modeled GPP showed higher variability, with an overall RMSE% of 43 % and 41 % against the MODIS and GOSIF datasets, respectively. The 3D-CMCC-FEM model has consistently demonstrated reliability across multiple data sources and spatial scales, establishing it as a robust tool for forest monitoring, being, capable of delivering insights at daily, monthly, and annual resolutions over broad and heterogeneous areas. This approach offers innovative and promising improvements in the continuity of forest data, supporting more informed decision-making in climate policy and environmental management
Robot-assisted pyeloplasty with direct pyelo-ureteral anastomosis for retrocaval ureter
No full textn
AI Act e trattamento di dati personali di natura particolare
No full textThe article analyses the European Union Regulation 2024/1689 on artificial
intelligence, with particular attention to the fundamental choices
and methods adopted by the European legislator in establishing a «harmonised
» legal framework, both in terms of substantive provisions and
with regard to the regulatory apparatus designed to ensure the functioning
of the complex organisational system of protection introduced
by the Regulation. The author further provides an in-depth examination
of the key legal definitions of «AI system» and risk, with a particular
focus on the potential graduations of «high risk» and «unacceptable»
high risk
SOMIGLIANZE, DIRITTI E DOVERI DELL’INTERPRETAZIONE UMBERTO ECO TRA STORIA DELLE IDEE E SEMIOTICA
No full textThis article tackles some of Umberto Eco’s ideas about normality and pathologies of interpretation by showing why he recovers certain paradigms in the history of ideas, particularly those
understood as encyclopedic sets of cultural habits depicting how things tend to go. While arising within concrete contexts, the theory and practice of interpretation, with their leading rules,
do not belong to the plane of the texts that are their objects. While Eco applied his theory of
controlled multi-interpretability of texts ever since the studies conducted in Opera Aperta, he
employed it as a test bed for hermeneutic paradigms, such as hermetic semiosis and the esoteric
interpretations of Dante, only after the semiotic turn. The article’s conclusion reflects upon the
lawfulness, if not the necessity, of a distinction between literary text and document tout court,
the duties of the interpreter, and achieving the rights of the interpretation
Development and validation of a computational tool to predict treatment outcomes in cells from high‐grade serous ovarian cancer patients
Get PDFTreatment of High-grade serous ovarian cancer (HGSOC) is often ineffective due to
frequent late-stage diagnosis and development of resistance to therapy. Timely selec-
tion of the most effective (combination of) drug(s) for each patient would improve
outcomes, however the tools currently available to clinicians are poorly suited to the
task. We here present a computational simulator capable of recapitulating cell
response to treatment in ovarian cancer. The technical development of the in silico
framework is described, together with its validation on both cell lines and patient-
derived laboratory models. A calibration procedure to identify the parameters that
best recapitulate each patient's response is also presented. Our results support the
use of this tool in preclinical research, to provide relevant insights into HGSOC
behavior and progression. They also provide a proof of concept for its use as a per-
sonalized medicine tool and support disease monitoring and treatment selection