A large amount of data collected in the social sciences are counts crossclassified into categories. These counts are non-negative integers and require special methods of analysis to model appropriately; log-linear models are one sophisticated method. The counts are modeled by the Poisson distribution, and related to the classifying variables through a logarithm. Models can then be built, critically analysed, evaluated and compared to develop a suitable statistical model for modeling the count data. Log-linear models are powerful enough to cope with many classifying variables, and permit many model building ideas similar to those in standard statistical regression.
Trauma-informed practice (TIP) and culturally responsive pedagogy (CRP) are frequently advanced as complementary equity approaches in schools, yet the practice-level interface between them remains under-specified—particularly in communities shaped by poverty, displacement, racism, and intergenerational trauma. This qualitative study examined how 26 primary school educators in a culturally diverse, socioeconomically disadvantaged Australian school described adapting trauma-informed practices to students' social and cultural contexts. Data were collected through semi-structured interviews and analysed using reflexive thematic analysis, with ecological systems theory providing an interpretive framework. Three themes captured educators' sense-making: (1) Prioritising safety and belonging, reflecting microsystem practices addressing basic needs shaped by socioeconomic adversity; (2) Navigating cultural complexity, illuminating mesosystem negotiations between institutional expectations and family cultural practices; and (3) Seeking deeper understanding, capturing educators' recognition of exosystem gaps between available professional learning and the intersecting realities of trauma, culture, and disadvantage. Findings highlight the interpretive labour educators undertake when adapting standardised frameworks to local contexts, while revealing tensions between individualised trauma responses and structural inequities. Implications for culturally responsive professional development, institutional policy reform, and educational practice are discussed.
With the continuous growth in the number of parameters of transformer-based pretrained language models (PLMs), particularly the emergence of large language models (LLMs) with billions of parameters, many natural language processing (NLP) tasks have demonstrated remarkable success. However, the enormous size and computational demands of these models pose significant challenges for adapting them to specific downstream tasks, especially in environments with limited computational resources. Parameter Efficient Fine-Tuning (PEFT) offers an effective solution by reducing the number of fine-tuning parameters and memory usage while achieving comparable performance to full fine-tuning. The demands for fine-tuning PLMs, especially LLMs, have led to a surge in the development of PEFT methods, as depicted in Fig. 1. In this paper, we present a comprehensive and systematic review of PEFT methods for PLMs. We summarize these PEFT methods, discuss their applications, and outline future directions. Furthermore, we conduct experiments using several representative PEFT methods to better understand their effectiveness in parameter efficiency and memory efficiency. By offering insights into the latest advancements and practical applications, this survey serves as an invaluable resource for researchers and practitioners seeking to navigate the challenges and opportunities presented by PEFT in the context of PLMs.
Climate change is accelerated by increasing levels of greenhouse gases (GHGs) as a result of human activity, particularly the release of carbon dioxide (CO2). Soil carbon (C) sequestration, or the transfer of atmospheric CO2 to soil organic matter (SOM) with long-term stabilization within the soil, is an important process of C removal from the atmosphere. For the accounting of soil C and offset markets in most countries including Australia, the standard soil sampling depth is 0–30 cm, although deeper sampling is recommended for more accurate C stock assessments and to capture long-term sequestration potential. While 30 cm soil depth accounts for most short-term management impacts on C storage, a significant portion of soil C is stored below this depth (i.e., deep soil C), and sampling at greater depths can provide a more complete account of total C stocks and potential sequestration benefits. This paper aims to provide a comprehensive review, including a bibliometric analysis and a critical discussion of the link between deep soil C storage and sequestration potential in relation to climate change mitigation and soil health. Deep soil layers contain over 850 Pg C worldwide, which is approximately 50
Compound Floods (CompF), driven by interactions among coastal, fluvial, and pluvial drivers, pose heightened risks to coastal cities under climate change (CC) and rapid urbanization. This study systematically reviews the causes, mechanisms, and assessment frameworks of CompF, with a particular focus on the role of AI in enhancing flood prediction and risk management. Following PRISMA guidelines, a systematic literature review was conducted using Web of Science and Scopus databases, covering 2015–2025. A total of 898 unique articles were analyzed through bibliometric, scientometric, and thematic analyses. Keyword trends, co-authorship, and country-level contributions were analyzed using VOSviewer, while thematic synthesis was employed to identify recurring patterns, gaps, and emerging research themes. These themes centered on the definitions, causes, mechanisms, forms, and models used to assess compound flooding in coastal cities. The impact of CC and the future projections are also documented. Finally, a conceptual framework is developed to integrate flood typology, drivers, assessment models, and CC projections. The review highlights a growing scholarly attention to CompF. Physical and statistical models remain dominant, while AI-driven and hybrid approaches are emerging, with climate change projections indicating more frequent and severe compound flood events. This study provides policymakers, planners, and scientists with an integrated framework for improved prediction, mitigation, and resilience strategies. It emphasizes the value of AI-based approaches and interdisciplinary collaboration in addressing future CompF risks in coastal cities.
