48080 research outputs found
Sort by
Development and Clinical Implementation of Detector Systems Based on 2D Monolithic Pixelated High-Spatial-Resolution Semiconductor Sensors for Dosimetry in Radiation Oncology
No full textModern radiation therapy is increasingly characterized by complex, non-uniform dose distributions with steep dose gradients, delivered dynamically through high-precision dose placement using small radiation fields, demanding stringent verification methods. Effective dosimetry under these conditions requires detectors with high spatial and temporal resolution, real-time readout capabilities, tissue equivalence, and minimal dependence on dose, dose rate, beam energy and direction.Silicon semiconductor detectors show great potential in this area. They enable the production of small sensitive volumes, including one-dimensional and two-dimensional arrays, without compromising mechanical robustness or sensitivity, thus achieving high spatial resolution. As natural dose-rate meters, they provide real-time response and time-resolved measurements. Additionally, the electron stopping power ratio of silicon to water remains almost constant within the secondary electron energy range relevant to X-ray radiotherapy, rendering the detectors water-equivalent. Silicon detectors thus possess many characteristics of an ideal dosimeter, while their known limitations – such as accumulated dose, dose rate, energy, temperature, and angular dependencies – can be compensated for or corrected and therefore minimized.This thesis focuses on the development and clinical implementation of detector systems based on two-dimensional monolithic pixelated silicon sensors of MagicPlate family for dosimetry in advanced X-ray radiotherapy modalities, particularly the MagicPlate-976 detector. MagicPlate-976, featuring 976 ion-implanted diodes (pixels) arranged non-uniformly over a 58x58 mm² area, achieves spatial resolution down to 1 mm centrally. Its readout electronics provide temporal resolution as fine as 0.2 ms without requiring external triggering.</p
Food Relief in Focus: A Snapshot of Food Relief in the Illawarra and Shoalhaven – 2025
No full textThe 2025 Food Relief and Rescue Survey collected responses from 15 organisations delivering food relief across the Illawarra–Shoalhaven. These organisations perform two critical functions: providing food to individuals and families in need and recovering surplus food that would otherwise be wasted. This combination of social and environmental benefit means food relief providers are an important part of the local community support network and the regional food system.Collectively, the organisations reported an estimated weekly impact of:1,540 meals served652 hampers and food parcels delivered1,723 individuals and 495 families supported</p
Single-molecule studies of replication-transcription conflicts and recombinational DNA repair
No full textDNA replication, transcription, and recombinational repair are essential processes that maintain the genomic integrity of bacterial cells by ensuring high-fidelity DNA duplication, accurate gene expression, and efficient DNA repair, thereby preventing the loss of genetic information. Multiple proteins execute these complex cellular processes on the DNA within the confined physical space of the bacterial cytoplasm. Among these, DNA replication is the fundamental process that underpins all other genomic maintenance mechanisms. It ensures the accurate duplication of genetic material, which is then faithfully passed on to daughter cells during cell division. Without high-fidelity replication, errors could accumulate in the genetic code, leading to mutations that compromise cellular functions. Additionally, replication provides a reliable template for transcription, which is essential for accurate gene expression. However, conflicts can arise when the replication fork encounters the transcription machinery, as both processes occur on the same DNA template, potentially stalling both and threatening genomic stability. These conflicts occur in two forms: co-directional conflicts, which occur when replication and transcription proceed in the same direction, and head-on conflicts, which arise when they move in opposite directions and collide. If unresolved, these conflicts can lead to mutations that contribute to the development of antibiotic resistance in bacteria and cancer and neurodegenerative disorders in humans.Earlier studies on replication-transcription conflicts have employed genetic, imaging, and biochemical approaches, revealing that head-on conflicts are more detrimental than co-directional ones, often leading to replication fork stalling, collapse, and DNA breakage. However, the exact outcomes of these conflicts and the mechanisms by which they are resolved remain poorly understood. Notably, since the early 1980s, when Bruce Alberts' laboratory first investigated these conflicts through ensemble studies using a reconstituted T4 replisome and E. coli RNA polymerase, no single-molecule techniques have been employed to examine their outcomes. This has left a critical gap in understanding the precise molecular events in real time. Single-molecule assays are best suited to examine this, as they allow direct observation of dynamic, transient events in real time, which are often masked in bulk measurements.To address this gap, the primary goal of this thesis is to develop and apply single-molecule tools to investigate the outcomes of replication-transcription conflicts. First, I established a novel method for constructing large circular rolling-circle DNA substrates using nCas9-gRNA complexes. These substrates are easily customizable and offer improved spatial resolution, making them ideal for single-molecule studies of interactions between molecular complexes involved in replication and transcription. Second, this thesis details the development of in vitro single-molecule fluorescence assays to visualize collisions between replication and transcription machinery. For the first time, I observed collisions between the E. coli replisome and RNA polymerase on large rolling-circle DNA substrates at the single-molecule level, outside the context of the cell. The results show that replication slows and eventually stalls when encountering RNA polymerase in a co-directional orientation, while head-on collisions lead to complete replication stalling. Additionally, this assay can be adapted to introduce different replication roadblocks, providing a framework for future investigations of replication-transcription conflicts in other systems, such as viruses and eukaryotes.In addition to protein roadblocks, bacterial DNA is constantly exposed to external environmental stressors such as UV radiation, harmful chemicals, and oxidative stress. These factors pose significant challenges to maintaining genomic integrity, as they can induce DNA lesions, including single and double-strand breaks, base modifications, and crosslinks. Replication first stalls upon encountering these lesions and often bypasses them, resulting in the formation of post-replication gaps. To preserve genomic integrity and ensure cell viability, it is crucial that cells efficiently detect, respond to, and repair these gap lesions while minimizing the risk of mutations. To counteract DNA gap lesions, bacterial cells rely on specialized, error-free recombination pathways involving RecA, which plays a key role in filling post-replication DNA gaps, with additional support from various factors of the RecFORJ and RecBCD pathways. However, the precise coordination of these recombination factors in repairing post-replication ssDNA gaps during recombinational repair after DNA damage remains unclear. Despite significant advancements in cell imaging technology in recent decades, single-molecule tools capable of directly observing post-replication gaps in live bacterial cells have been lacking.The final part of my thesis addresses this gap by employing an in vivo single-molecule live-cell imaging approach. Using fluorescence microscopy, I directly visualized the formation and resolution of SSB-bound ssDNA intermediates of various sizes in live wild-type and recombination mediator protein-deficient E. coli strains expressing fluorescent SSB-mTurquoise2 proteins following UV exposure. The results demonstrate that, in the presence of all recombination factors, cells exhibit fewer SSB-bound ssDNA features after DNA damage. In contrast, the formation and resolution of SSB-bound ssDNA intermediates within cells depend on specific recombination factors, highlighting the distinct roles of recombination proteins in resolving ssDNA intermediates that may represent post-replication gaps.Together, the results from both in vitro and in vivo single-molecule studies conducted in this thesis contribute to our understanding of the complex biological processes of RTCs and recombinational DNA gap repair mechanisms in E. coli and provide a foundation for future investigations, with potential applications in viral and eukaryotic systems.</p
nCas9-based method for rolling-circle DNA substrate generation
No full textRolling-circle DNA replication is a DNA-duplication mechanism whereby circular DNA templates are continuously copied to produce long DNA products. It is widely used in molecular diagnostics, DNA sequencing, nanotechnology, and in vitro DNA replication studies. The efficiency of rolling-circle replication reaction heavily relies on the quality of the rolling-circle DNA template. Existing methods to create rolling-circle DNA substrates often rely on unique restriction sites and have limited control over replication fork topology and position. To address these limitations, we present a straightforward, customizable, and efficient strategy for producing rolling-circle DNA substrates with control over gap size and fork position. Our method relies on the use of nickase Cas9 (nCas9), which can be programmed to target specific DNA sequences using guide RNAs. In a one-pot reaction, we target nCas9 to four sites on an 18-kb plasmid to create 8–11-bp fragments. These fragments are removed and a flap oligo is ligated, to construct a fork with precisely controlled flap length and gap size. We demonstrate the application of this DNA substrate in an in vitro single-molecule rolling-circle DNA-replication assay. With our method, any plasmid DNA can be converted into a rolling-circle template, permitting generation of more physiologically-relevant DNA templates.</p
Wearable sensors for continuous, real-time health and safety monitoring and risk assessment
No full textMine workers are continuously exposed to a host of non-fatal factors, such as, heat and high humidity that can pose a serious risk to their health and wellbeing either through direct effects, such as, heat exhaustion, or indirect effects, such as, degraded decision-making ability. Wearable sensors and contextual analytics provide a platform to unobtrusively collect and fuse multimodal health and safety data in real-time to compute both acute safety and longitudinal health insights. These data can also be used to assist in identifying the risk level associated with each worker and their planned activity. Preliminary trials of wearable sensors with mine workers revealed the acceptability of such equipment and showed the potential benefits of using wearable sensors for gathering health related biometric data. To further explore the data gathering and potential usage of wearable sensors in a mining related environment, testing with recruits undergoing training with the Queensland Mines Rescue Service has been undertaken. Quantitative (biometric) and qualitative data was gathered from the participants to determine the effects of various training scenarios and equipment on heat stress indicators. The results show that both heart rate and core temperature have a positive correlation with task difficulty.</p
Modelling warfare between vampires and humans: who wins?
No full textWe adapt the Lanchester combat model to represent conflict between vampires and humans. It is assumed that vampires attack humans during the hours of darkness whilst humans attack vampires during the hours of light. The right-hand side of the differential equation model therefore depends upon the hour of the day. A key insight is that to answer the question ‘who wins’ it is not required to ‘stitch together’ the solutions to the differential equation over many days. Rather, the number of combatants surviving at the end of each day can be cast as a difference equation in terms of the numbers surviving at the end of the previous day. Two models are investigated and from the solutions the boundary delimiting the parameter regions of victory and defeat is found. Where appropriate we introduce discussion points, both mathematical and modelling. The mathematical techniques used include finding the eigenvalues and eigenvectors of a matrix, solving linear differential and discrete systems, and considering the physical meaning of the model and questioning its assumptions. Thus a variety of problem-solving skills are developed within the context of a model where students can question the underlying assumptions and propose, and investigate, models of their own.</p
Early Childhood Education Movement Environments: An Exploration of Quality, Children’s Outcomes, and the Potential Efficacy of Professional Development
No full textBackground Holistic child growth and development is underpinned by five interconnected child development domains: the social, emotional, cognitive, language and physical domains. The physical domain is comprehensive and comprises elements related to movement, reflexes, sensors, physical development and physical activity. Everyday movements and interactions support optimal growth and development in the physical domain, and thus, deficiencies in this domain can negatively impact cognitive development, language development, social behaviours, peer relationships and self-esteem.Early childhood education (ECE) settings have a critical role in children's physical development. Thus, physical development is a key component of national and international ECE curricula, as well as national and international policy and practice documents. Several ECE factors have been identified as important in the promotion of physical development in ECE settings, such as intentional teaching opportunities and the quality of the ECE environment. The quality of the ECE environment is directly related to children’s outcomes, that is, higher quality ECE environments result in more favourable health and well-being outcomes for children. The quality of the ECE environment that relates to the physical child development domain is known as the movement environment. To date, few studies have explored aspects relating to the quality of the movement environment, with the MOVERS scale recently being developed to assess the ECE movement environment. Therefore, the aims of this doctoral thesis were to: (1) determine the validity and reliability of the MOVERS scale (Chapter 3, paper 1), (2) explore the relationships between the quality of the movement environment and children’s physical activity, gross motor skills and fine motor skills (Chapter 4, paper 2), (3) explore the relationship between the quality of the movement environment and children’s academic outcomes (numeracy and literacy) (Chapter 5, paper 3), (4) determine the potential efficacy of the MOVERS Professional Development Program on the quality of the movement environment and children’s outcomes (Chapter 6, paper 4).Methods The validity and reliability of the MOVERS scale were determined using data from 13 ECE settings. Standardised tests, such as ICCs, percentage agreement and weighted Kappa, Cronbach’s alpha and correlation coefficients, were used to test the test-retest reliability, internal consistency and concurrent validity (Chapter 3). Linear and generalised mixed models were used to determine the relationship between the quality of the movement environment and children’s physical activity, gross and fine motor skills and numeracy and literacy skills (Chapters 4 and 5). In Chapter 6, linear mixed models were used to determine the potential efficacy of the MOVERS Professional Development Program on the quality of the environment and children’s outcomes. In each study, the quality of the movement environment was assessed using the MOVERS scale and objective measures of physical activity (accelerometers), gross (TGMD-2) and fine motor skills (ASQ-3), numeracy (PENS) and literacy (PPVT-4) were used in all studies.Results Chapter 3 reported the validity and reliability of the MOVERS Scale. The MOVERS scale was shown to have excellent test-retest reliability and excellent internal consistency. The investigation of the concurrent validity of the MOVERS showed mixed findings. Chapter 4 reported positive relationships between the quality of the movement environment and children’s physical activity and gross motor skills. No significant relationships were found between the quality of the ECE movement environment and children’s cognitive outcomes (Chapter 5). In Chapter 6, large effect sizes were reported for the quality of the ECE movement environment. Medium to large effect sizes were reported for children’s gross motor skills, and small to medium effect sizes were reported for numeracy skills. In the intervention group, a greater number of children moved from being at risk of delay or delayed to developmentally on track in fine motor skills.Discussion and Conclusions This research program significantly contributes to the current evidence pertaining to the quality of ECE movement environments. Researchers can be confident that the MOVERS scale is a reliable scale to assess the quality of the movement environment in ECE settings. Further research regarding the concurrent validity of the MOVERS scale would be valuable. This research also explored relationships between the quality of the movement environment and several child-related outcomes (Chapters 4 and 5). The positive relationships between the quality of the movement environment and children’s physical activity, gross and fine motor skills and numeracy skills are important. It is feasible to suggest that further relationships may be present in larger sample sizes with increased heterogeneity in the quality of the movement environment. As demonstrated in Chapter 6, educator professional development is a potentially effective method of improving the quality of the movement environment, as well as improving children’s outcomes. Further professional development opportunities focusing on the physical child development domain and the quality of the movement environment are needed within the Australian context and beyond.This program of research addressed several significant gaps in the current literature pertaining to the quality of the movement environment in ECE settings and the child physical development domain. Supporting and promoting optimal physical development from a young age is critically important for children’s health and well-being in the short and long term. Continued advocacy is needed within the ECE sector to continue this important work. The results from this research will provide a strong foundation for future research.</p
Modelling Augmentation of Marketing Ambidexterity, Competitive Advantage and Sustained Profitability of Firms through Artificial Intelligence (AI)-Powered Customer Relationship Management (CRM) Capabilities
No full textArtificial Intelligence (AI) has become a transformative force within Customer Relationship Management (CRM), as it enables organisations to revolutionise their customer interactions, personalise their marketing strategies, and streamline their operational efficiency. AI-powered CRM systems, such as Salesforce, Zoho, and HubSpot, are being increasingly adopted by businesses to enhance their productivity and performance. Despite this trend, a significant gap remains in our understanding of the specific dimensions of AI-powered CRM’s capabilities and how these contribute to organisational outcomes such as sustained profitability and sustained competitive advantage. This study addresses these gaps by exploring two central research questions: (1) What are the dimensions and sub-dimensions of AI-powered CRM capabilities? and (2) How do these capabilities, particularly through the mechanism of marketing ambidexterity, influence sustained profitability and sustained competitive advantage of firms?This research adopts a robust, multi-method approach, comprising three distinct studies. Study-1 involves a scoping literature review that systematically maps the existing academic and industry landscape, providing a foundational understanding of AI-powered CRM systems. Study-2 consists of a qualitative analysis based on 24 in-depth interviews with AI-powered CRM system experts, which identifies three major dimensions of AI-powered CRM: data management capability, multi-channel integration capability and service offerings capability. These dimensions are further divided into sub-dimensions; namely, data governance, data intelligence, data privacy, content consistency, process consistency, personalisation, automation and meaningfulness. The findings of these studies are synthesised into a hierarchical conceptual model that elucidates the relationships between AI-powered capabilities and key organisational outcomes.To validate this model empirically, Study-3 was conducted, entailing a quantitative survey of 205 employees from leading banks, the results of which were analysed using Partial Least Squares (PLS) path modeling. The results of Study-3 confirm that AI-powered CRM capabilities significantly enhance both sustained profitability and sustained competitive advantage, with marketing ambidexterity serving as an important mediator. This mediation highlights the importance of balancing exploration and exploitation within marketing strategies in order to harness fully the potential of AI-powered CRM systems.</p
Terahertz Vibrations and Thermal Conductivity in Nanodiamond-Embedded Thermoelectric Composites
No full textWith globally increasing energy demands and the desire for efficient and sustainable energy production, the ability to generate electricity lfom an energy source that occurs naturally and is a byproduct of other processes is an extremely attractive prospect. Thermal energy, or heat, is one such source. Among the various technologies capable of converting heat into electricity, thermoelectric (TE) devices offer a solid-state solution that is scalable, reliable, and maintenance-free. Currently, TE devices have limited applicability due to their relatively low energy conversion efficiencies. Enhancing efficiency requires strategic material design, guided by the "phonon-glass electron-crystal" concept introduced by Slack in 1995. This approach emphasises the need for TE materials that efficiently conduct electricity but do not efficiently conduct thermal energy.This thesis focuses on the thermal conductivity side of the TE material design problem while aiming to provide insights into thermal transport in general, which is important for energy management beyond TE-based applications. Carbon nanoparticles (NPs) are investigated for their ability to alter the thermal transport properties of a host material when composited. Due to their unique geometric, structural, and thermal properties, carbon NPs such as carbon nanotubes, graphene NPs, and nanodiamonds have been studied as additives to many different materials. In TE materials, where heat is primarily carried by THz frequency atomic vibrations, NPs have been shown to strongly reduce thermal transport while having minimal effects on electrical transport. However, a detailed understanding of the thermal transport mechanisms in nanocomposite materials is lacking, making it difficult to engineer thermal conductivity in TE and other materials optimally.One TE-NP composite system was studied in detail in this thesis to provide insights into the thermal conductivity mechanisms in TE-NP composites and produce a novel, high-performance TE material. For the TE phase, tin telluride (SnTe) was chosen. SnTe shows promising TE performance at temperatures relevant to automotive and industrial applications (~300-800 K), without the environmental cost associated with similar materials like PbTe, and is also a crystalline topological insulator, which gives it additional technological interest. For the carbon NP additive, nanodiamond (ND) particles (~5 nm diameter) were chosen due to their high dispersibility, matching cubic crystal structure, high surface area to volume ratio, and relative novelty as a TE additive.In the pursuit of engineering thermal transport in TE-NP composites, another, more fundamental goal was invoked: to provide deeper insights into the THz-scale vibrational properties of NDs themselves. Studying the vibrational properties of NDs has two important benefits: 1) NDs are of technological interest for applications in sensing (e.g., nanoscale thermometry and biological imaging) and quantum information, which are influenced by their vibrational properties, and 2) to understand the unique nano-scale vibrational properties of NPs, it is helpful that the bulk phase is already well-studied and understood, as is the case with diamond.SnTe-ND composites were synthesised using the spark plasma sintering technique and found to exhibit enhanced TE properties over pure SnTe when the ND concentration was ≤ 1 .0 vol.%. This is because the SnTe- ND interface does not allow for effective thermal transfer due to the large mismatch in acoustic properties between the materials. Meanwhile, electronic transport is unhindered due to the robustness of electronic states in SnTe which can transport electrons easily around the diamond particles. By measuring the THz phonon structure of SnTe via inelastic neutron scattering and comparing it with ND-composited SnTe, changes in phonon linewidths are observed, indicating reduced phonon lifetimes due to scattering caused by the NDs. However, the changes are subtle, especially when considering the drastic changes in thermal conductivity measured. In fact, the temperature dependence and plateau value of the lattice thermal conductivity at high temperatures suggest that, when composited with a small percentage of NDs, SnTe goes from being well-described by particle-like phonon-mediated thermal transport to wave-like diffuson-mediated thermal transport—a new paradigm for nanocomposites that is adopted from glassy materials.To further understand the thermal transport in the SnTe-ND system and TE-NP systems more generally, vibrational studies of NDs were undertaken using a combination of classical molecular dynamics simulations and inelastic neutron scattering measurements. Compared with bulk diamond, a range of unique vibrational dynamics is revealed in NDs due to their finite size and surfaces, including THz surface phonons, low-energy whole-particle resonances (Lamb modes), a boson-like peak, surface rattling modes, and soft and broad core acoustic and optical phonons. Of particular interest is the observation of Rayleigh surface phonons, which have a low group velocity and an amplitude that decays exponentially away from the surface. These Rayleigh surface phonons, which are like ultrafast, ultra-large (with respect to object diameter) earthquake waves, are responsible for the previously observed but unexplained linear scaling of the low-energy vibrational density of states in NPs. The unique vibrational properties observed in NDs are likely to impact their intrinsic thermal transport and interfacial thermal transport and may have important implications for their applications in sensing and quantum information. Additionally, many of these properties are likely to be universal to NPs, influencing their thermodynamic properties and having implications for other NP applications like catalysis.The nature of ND lattice vibrations was lastly studied in the context of embedded particles in the SnTe matrix. Time-of-flight neutron spectroscopy resolved phonons from a small concentration of elementally light NPs when uniformly embedded in a heavy matrix for all possible excitations between ~0.5-250 meV—an extremely difficult challenge for other (e.g., optical) techniques. To provide a full picture of the influences of matrix embedding on ND phonons, the effects of temperature and local particle environment were also studied. Upon embedding, the quenching of surface phonons and the softening and narrowing of core phonons due to the new boundary conditions and tensile strain imposed by the matrix were observed. Additionally, the restriction of the ND surface limits phonon softening with increasing temperature and narrows the DOS features which were broadened by the free anharmonic particle surfaces. These changes in vibrational properties upon embedding will a) influence their intrinsic, and thus composite, thermal properties, and b) can be used to make observations about the nature of the local chemical environment of the ND. For example, subtle differences in the inelastic neutron signal can be observed from composite samples with many agglomerates, since agglomerated NDs have slightly more freedom than completely embedded NDs.Overall, this thesis: 1) highlights the need to incorporate diffuson-mediated thermal transport into nanocomposite thermal conductivity models, 2) highlights the effectiveness of NDs as additives for reducing lattice thermal conductivity in TEs, 3) presents a novel, high-performing SnTe-ND composite material with ultra-low lattice thermal conductivity, 4) provides a comprehensive framework for the vibrational properties of isolated and embedded NDs, and 5) demonstrates the utility of neutron spectroscopy for studying the THz dynamics of NPs and nanocomposite systems, especially when combined with atomistic simulations.</p
Multi-Objective Optimisation of Sequential Recommender Systems through Advanced User Preference Modelling
No full textSequential recommendation, as a branch of recommender systems, has been widely developed due to its ability to capture behavioural sequences and dynamic preferences. By analysing past user interactions, sequential recommender systems aim to predict current preferences and provide recommendations that adapt to users’ evolving needs. However, existing sequential recommendation models often fall short of accurately modelling users’ real-time preferences due to the complexity and diversity of user interests. Many current approaches rely on simplified modelling techniques that struggle to capture the subtle variations in user behaviour and content features. Moreover, focusing solely on improving recommendation accuracy can reduce the overall effectiveness of the system, as other crucial objectives, such as long-tail item inclusion and recommendation diversity, remain unaddressed. A strong bias towards popular items reduces the visibility of niche content, restricting user discovery and engagement. Additionally, rigid user preference modelling can lead to homogeneous recommendations, making the user experience repetitive and less engaging.These limitations indicate that existing sequential recommendation models struggle to balance accuracy, diversity, and fairness simultaneously. Addressing these gaps requires a more adaptive and multi-objective approach that can refine user preferences dynamically, integrate auxiliary information effectively, and incorporate trust-based mechanisms to enhance recommendation diversity.Along these lines, this thesis aims to tackle these issues and achieve multi-objective optimisation in sequential recommendation through an advanced analysis of user preferences. The primary objectives include enhancing recommendation accuracy across different usage scenarios, mitigating the long-tail effect, and improving recommendation diversity to create a more balanced and engaging user experience.Firstly, to improve sequential recommendation accuracy, we propose a novel model that leverages item features for more sophisticated user preference modelling. This model integrates item features into a knowledge graph, allowing for a richer representation of user preferences beyond simple interaction-based approaches. Additionally, by framing the recommendation process as a reinforcement learning (Q-Learning) problem, the system can continuously adapt to users’ evolving interests, leading to improved recommendation accuracy. We evaluated this model on publicly available datasets, and experimental results confirm its effectiveness in enhancing user satisfaction, particularly in interactive recommendation scenarios.Secondly, to ensure robust performance in cases where item features are unavailable, we propose an alternative modelling approach based on interaction time intervals between users and the system. By incorporating temporal information, this method captures preference shifts in real time, reflecting dynamic user behaviours more effectively. Experimental evaluations on publicly available datasets demonstrate that this approach significantly outperforms conventional models that do not consider time intervals, achieving greater recommendation accuracy.Thirdly, to mitigate the long-tail effect in sequential recommendations, we propose a feature-level preference modelling approach. This method enhances user preference representation by integrating structured item attributes, allowing for more precise preference alignment. By combining users’ sequential interaction patterns with feature-driven matching, the system increases the visibility of relevant long-tail items that might otherwise receive limited exposure. Experimental validation on public datasets demonstrates that this approach effectively mitigates the long-tail issue while also improving overall recommendation accuracy.Lastly, to enhance recommendation diversity, we develop a sequential recommendation model that incorporates trust-based mechanisms. We first propose a trust model that quantifies users’ trust levels based on both explicit and implicit trust relationships, leveraging user-generated content to enrich preference estimation. This trust model is then integrated into the sequential recommender system to improve diversity-aware recommendations. By considering the preferences of users’ trusted connections, the model broadens recommendation scope while minimising accuracy loss. Experimental results on public datasets confirm that this approach significantly improves recommendation diversity while maintaining high accuracy, ensuring a more engaging and varied recommendation experience.</p