1,723,880 research outputs found
The impact of consolidating web based social networks on trust metrics and expert recommendation systems
No full textIndividuals are typically members of a variety of web-based social networks (both explicit and implied), but existing trust inference mechanisms typically draw on only a singlenetwork to calculate trust between any two individuals. This reduces both the likelihood that a trust value can be calculated (as both people have to be members of the samenetwork), and the quality of any trust inference that can be drawn (as it will be based on only a single network, typically representing a single type of relationship). To make trust calculations on Multiple Distributed (MuDi) social networks, those networks must first be consolidated into a single network.Two challenges that arise when consolidating MuDi networks are their heterogeneity, due to different name representation techniques used for participants, and the variability of trust information, due to the different trust evaluation criteria, across the different candidate networks. Semantic technologies are vital to deal with the heterogeneity issues as they permit data to be linked from multiple resources and help them to be modelled in a uniform representation using ontologies. The inconsistency of multiple trust values from different networks is handled using data fusion techniques, as simpler aggregationtechniques of summation and weighted averages tend to distort trust data.To test the proposed semantic framework, two set of experiments were run. Simulation experiments generated pairs of networks with varying percentages of Participant Overlap (PO) and Tie Overlap (TO), with trust values added to the links between participants in the networks. It analysed different data fusion techniques aiming to identify which best preserved the integrity of trust from each individual network with varying values of PO and TO. A real world experiment used the findings of the simulation experiment on the best trust aggregation techniques and applied the framework to real trust data between participants that was extracted from a pair of professional social networks. The trust values generated from consolidated MuDi networks were then compared with thereal life trust between users, collected using a survey, with the aim of analysing whether aggregated trust is closer to real life trust than using each of the individual networks.Analysis of the simulation experiment showed that the Weighted Ordered Weighted Averaging (WOWA) data fusion technique better aggregated trust data and, unlike theother techniques, preserved the integrity of trust from each individual network for varying PO and TO (p ? 0.05). The real world experiment partially proved the hypothesis ofgenerating better trust values from consolidated MuDi networks and showed improved results for participants who are part of both networks (p ? 0.05), while disproving theclaim for those in the cross-region (with one user present in both networks and the other in a single network) and single-network users (p > 0.05)
Synthesis and Post-synthesis Transformations of Colloidal Semiconductor Nanocrystals
No full textThe present PhD thesis focuses on two main classes of semiconductor colloidal nanocrystals, i.e. lead halide perovskite and copper chalcogenides. The former class of semiconductor NCs are promising materials for many high performance optoelectronics applications, as they exhibit a tunable band gap in the range of 1.4 to 2.9 eV and an efficient photoluminescence characterized by narrow emission linewidths and have been explored the most in the last years. Following the standard hot injection based synthesis and selecting a combination of short chain acid (octanoic acid or hexanoic acid) together with alkyl amines (octylamine and oleylamine) we prepared strongly fluorescent CsPbBr3 perovskite nanowires with tuneable width, in the range from 20 nm (exhibiting no quantum confinement, hence emitting in the green) to 3 nm (in the strong quantum-confinement regime, emitting in the blue) for the first time. However the main limitation of the colloidal synthesis protocols that was followed in aforementioned case including the ligand assisted reprecipitation routes which is the second most frequently used method for preparation of LHPs, is that they employ PbX2 (X= Cl, Br, or I) salts as both lead and halide precursors which consequently limit the precise tunability of the amount of reaction species such as metals or halides precursors and are not applicable to entire family of APbX3 (A=FA, MA and Cs; X=Cl, Br, I). To overcome this issue we developed benzoyl halide based colloidal synthesis route i.e broadly applicable to the entire family of LHP NCs and not only ensures the independent tunability of reaction precursors but also maintain the overall integrity of the NCs such as phase purity and high PLQY. Despite the significant advances in synthesis procedures, the control over size monodispersity, shape and phase purity remains another long standing challenge. This is in fact due to the tendency of primary alkyl amine in the form of alkylammonium ions that could compete with Cs+ ions and leads to the anisotropic growth such as NPLs or their use in excess permotes the Pb-depleted Cs4PbX6 phases. We develop here a strategy to achieve size, shape and phase pure CsPbBr3 nanocubes by substituting primary alkyl amines with secondary alkyl amines. We attributed this excellent control over the shape and phase purity to the inability of secondary amines to find the right steric conditions at the surface of the nanocrystals which consequently limits the formation of low dimensional structures. The shape purity and narrow size distribution leads to their ease of self-assembly in superlattices reaching up to 50 microns in lateral dimensions, which are the largest dimensions reported to date for superlattices of LHP NCs. The second class of materials studied here, i.e. copper chalcogenides, are mainly attractive due to their tunable composition via post synthesis chemical transformations, plasmonic properties, low toxicity and environmental friendliness. Taking the advantage of colloidal synthesis and using Cu2S as a template we develop a strategy to obtain novel AuCuS-Cu2S heterostructure through cation exchange, which cannot be realized through conventional synthesis approaches. We further investigated the stability of Cu2S NCs with different dimensionalities and their thermal evolution subsequent to the metal decoration. Interestingly the presence of additional metallic NCs, such as Au and Pt not only improves their thermal stability but also leads to the formation of bi-metallic alloys semiconductor heterostructure
Sull'automazione del rilevamento dei problemi di prestazioni nei sistemi software attraverso l'analisi empirica
No full textPerformance problems in software systems often go unnoticed until they degrade user experience or system trustworthiness. Detecting such issues requires targeted profiling or specialized tests, which are costly to develop, time-consuming to execute, and impractical to scale. As a result, developers often lack early feedback on the performance impact of their code changes. These limitations point to a broader challenge in software engineering. Developers often need to know what to test and where performance bugs might exist. They also need to understand how changes in the code may affect execution time and energy usage. Gathering these insights early and efficiently is difficult, especially without heavily relying on software behavior at runtime.
This thesis investigates whether it is possible to support early and scalable performance reasoning without relying on exhaustive profiling. We explore the use of static code features, lightweight dynamic features, and learning-based models to predict or explain performance-relevant properties at the method level. The goal is to assess how these approaches can help developers identify performance-sensitive code, understand testing needs, and estimate the potential impact of code changes on execution time and energy usage. The research is driven by the hypothesis that early and scalable performance reasoning is possible without relying on repeated execution or specialized instrumentation.
To evaluate this hypothesis, the thesis presents four empirical studies, each addressing a specific challenge in the software performance engineering (SPE) pipeline. The first study investigates the limited coverage of performance tests and reveals that performance tests are not only fewer than functional ones, but also more expensive to execute. The second study examines whether static code features can help to predict which methods are likely to be performance-tested, uncovering that such features offer limited predictive power across diverse software systems. The third study evaluates large language models (LLMs) for detecting performance bugs directly from source code. While LLMs show some promise, they struggle with reasoning on dynamic behavior. The final study focuses on energy consumption and investigates whether method-level energy usage can be predicted using static features and execution time, offering partial success but also highlighting modeling limitations.
Together, these studies map the opportunities and limitations of static analysis and machine learning in the detection of performance problems. The findings indicate that, while static features alone are insufficient, they may still contribute when combined with dynamic insights or advanced predictive models . The thesis contributes curated datasets, reproducible pipelines, and model artifacts to support future research in performance-aware and energy-aware software engineering
Reducing EMF via energy-efficient inter-frequency small cell discovery
No full textNo abstract available
Control data separation and its implications on backhaul networks
No full textFuture cellular systems need to cope with a huge amount of data and diverse service requirements in a flexible, sustainable, green and efficient way with minimal signalling overhead. This calls for network densification, a short length wireless link, efficient and proactive control signalling and the ability to switch off the power consuming devices when they are not in use. In this direction, the conventional alwayson service and worst-case design approach has been identified as the main source of inefficiency, and a paradigm shift towards adaptive and on-demand systems is seen as a promising solution. However, the conventional radio access network (RAN) architecture limits the achievable gains due to the tight coupling between network and data access points, which in turn imposes strict coverage and signalling requirements irrespective of the spatio-temporal service demand, channel conditions or mobility profiles
Comparison of Traffic Control with Model Predictive Control and Deep Reinforcement Learning
No full textTraffic congestion is among the worst causes of pollution, and the time spent in traffic can cost the world tens of billions of dollars every year. Solutions to mitigate this problem are at hand thanks to the advent of advanced control techniques and artificial intelligence (AI). Traditional traffic light control strategies based on fixed timing of the green, yellow and red phases are simple to implement, but at the same time very inefficient, in particular for busy intersections. This paper discusses both a model predictive control (MPC) approach and a model-free deep reinforcement learning (DRL) algorithm for controlling the traffic lights at a single intersection, with the aim of improving the traffic flow. Firstly, a detailed linear mathematical model of an intersection is formulated and successively tested in a MPC framework; secondly, a DRL algorithm is proposed and verified by comparing it with the currently implemented baseline controller. Finally, the results for the three approaches, MPC, DRL and the baseline controller, are validated through the SUMO (Simulation of Urban Mobility) microscopic traffic simulator
Evolving waste management: The impact of environmental technology, taxes, and carbon emissions on incineration in EU countries
Full text linkAmid the urgent global imperatives concerning climate change and resource preservation, our research delves into the critical domains of waste management and environmental sustainability within the European Union (EU), collecting data from 1990 to 2022. The Method of Moments Quantile Regression (MMQR) results reveal a resounding commitment among EU member states to diminish their reliance on incineration, which is evident through adopting green technologies and environmentally conscious taxation policies, aligning with the European Union's sustainability objectives. However, this transition presents the intricate task of harmonizing industrial emissions management with efficient waste disposal. Tailoring waste management strategies to accommodate diverse consumption patterns and unique circumstances within individual member states becomes imperative. Cointegrating regressions highlighted the long-run relationship among the selected variables, while Feasible Generalized Least Squares (FGLS) and Panel-Corrected Standard Errors (PCSE) estimates roughly confirmed MMQR results. ML analyses, conducted through two ensemble methods (Gradient Boosting, GB, and Extreme Gradient Boosting, XGBoost) shed light on the relative importance of the predictors: in particular, environmental taxation, consumption-based emissions, and production-based emissions greatly contribute to determining the variation of combustible renewables and waste. This study recommends that EU countries establish monitoring mechanisms to advance waste management and environmental sustainability through green technology adoption, enhance environmental taxation policies, and accelerate the renewable energy transition
- …
