Science Oxford is part of a charitable organisation called The Oxford Trust, based in Oxford, England. Science Oxford is the trust's education and engagement branch. The Oxford Trust was founded in 1985 by Sir Martin and Lady Audrey Wood. Science Oxford was founded in 2006. Science Oxford operates the Science Oxford Centre, a science museum located in Headington, and educational programmes.
BACKGROUND:Educational and skills-based interventions are often used to prevent relationship and dating violence among young people. OBJECTIVES:To assess the efficacy of educational and skills-based interventions designed to prevent relationship and dating violence in adolescents and young adults. SEARCH METHODS:We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, CINAHL, PsycINFO, six other databases and a trials register on 7 May 2012. We handsearched the references lists of key articles and two journals (Journal of Interpersonal Violence and Child Abuse and Neglect). We also contacted researchers in the field. SELECTION CRITERIA:Randomised, cluster-randomised and quasi-randomised studies comparing an educational or skills-based intervention to prevent relationship or dating violence among adolescents and young adults with a control. DATA COLLECTION AND ANALYSIS:Two review authors independently assessed study eligibility and risk of bias. For each study included in the meta-analysis, data were extracted independently by GF and one other review author (either CH, JN, SH or DS). We conducted meta-analyses for the following outcomes: episodes of relationship violence, behaviours, attitudes, knowledge and skills. MAIN RESULTS:We included 38 studies (15,903 participants) in this review, 18 of which were cluster-randomised trials (11,995 participants) and two were quasi-randomised trials (399 participants). We included 33 studies in the meta-analyses. We included eight studies (3405 participants) in the meta-analysis assessing episodes of relationship violence. There was substantial heterogeneity (I(2) = 57%) for this outcome. The risk ratio was 0.77 (95% confidence interval (CI) 0.53 to 1.13). We included 22 studies (5256 participants) in the meta-analysis assessing attitudes towards relationship violence. The standardised mean difference (SMD) was 0.06 (95% CI -0.01 to 0.15). We included four studies (887 participants) in the meta-analysis assessing behaviour related to relationship violence; the SMD was -0.07 (95% CI -0.31 to 0.16). We included 10 studies (6206 participants) in the meta-analysis assessing knowledge related to relationship violence; the results showed an increase in knowledge in favour of the intervention (SMD 0.44, 95% CI 0.28 to 0.60) but there was substantial heterogeneity (I(2) = 52%). We included seven studies (1369 participants) in the meta-analysis assessing skills related to relationship violence. The SMD was 0.03 (95% CI -0.11 to 0.17). None of the included studies assessed physical health, psychosocial health or adverse outcomes. Subgroup analyses showed no statistically significant differences by intervention setting or type of participants. The quality of evidence for all outcomes included in our meta-analysis was moderate due to an unclear risk of selection and detection bias and a high risk of performance bias in most studies. AUTHORS' CONCLUSIONS:Studies included in this review showed no evidence of effectiveness of interventions on episodes of relationship violence or on attitudes, behaviours and skills related to relationship violence. We found a small increase in knowledge but there was evidence of substantial heterogeneity among studies. Further studies with longer-term follow-up are required, and study authors should use standardised and validated measurement instruments to maximise comparability of results.
Reza Shah and Mohammad Reza Shah Pahlavi, father and son, ruled Iran between 1926 and 1979. During their reign Iran saw all seasons, including modernisation, dictatorship, arbitrary rule, chaos, foreign invasion and revolution. It was also a period in which the Pahlavis’ nationalist ideology clashed with democratic ideals, communist aspirations and – ultimately – Islamist beliefs. This article is an investigation in this regard.
Human activities have accelerated species extinctions, driving rapid biodiversity decline. Simultaneously, advancements in Artificial Intelligence (AI) and autonomous systems offer transformative potential for biodiversity research. Uncrewed vehicles—drones (aerial systems) and other robots (ground and underwater platforms)—equipped with high‐resolution sensors enhance ecosystem monitoring with unprecedented efficiency and scale. Ecologists must examine how these technological advancements can be used to monitor, understand and protect ecosystems more effectively. Here, we review studies published in Web of Science (1930–2023) using uncrewed vehicles for ecological monitoring and explore how these applications could be extended to further biodiversity research. We found drones dominate vegetation mapping, species monitoring and habitat assessment; underwater vehicles focus on benthic surveys and water quality monitoring; and ground robots are primarily used for sample collection. A greater use of drones reflects their commercial availability, lower costs and ease of deployment and management rather than technological maturity compared to other robots. Despite these applications, key gaps remain: research overwhelmingly targets plants (46%) and animals (44%), with minimal focus on microbes (10%) reflecting the limited use of uncrewed vehicles for specimen or environmental material sampling. Biodiversity hotspots, including South Africa, Central America and South America, are underrepresented, largely reflecting disparities in national research and economic capacity. Moreover, although uncrewed vehicles are technically capable of sample collection and real‐time surveillance, these functions remain underexploited—representing a promising avenue to unlock their full potential in biodiversity research. To maximise the impact of these technologies, taxonomic and biogeographic coverage must expand alongside functional applications. We argue that integrating uncrewed vehicles, advanced payloads and AI‐driven real‐time data analysis through collaborations among ecologists, roboticists and AI researchers will enable cost‐effective, autonomous ecological monitoring, advancing biodiversity conservation and addressing pressing knowledge gaps in the Anthropocene.
This study demonstrates an alignment of per-word processing time in a popular state-space language model Mamba and human readers. In Mamba, the recurrent state transition at each layer conceptually takes some duration of time, the discretization timestep Δ_t, determined dynamically in response to the input. Using a naturalistic reading dataset, we show that the per-word timestep from Mamba is a significant predictor of human reading times, and remains significant even when known predictors such as GPT-2 surprisal are controlled for. We further suggest, through formal analysis of Mamba's architecture and internal dynamics, that Mamba can serve as a new, valuable lens to look at human real-time language processing with ever-updated memory, because it allows us to look at how each module (layer) weighs short- and long-term information retention, and how noise may interact with dynamic, continuous memory representation. Code is available online.
Public spaces such as transport hubs, city centres, and event venues require timely and reliable detection of potentially violent behaviour to support public safety. While automated video analysis has made significant progress, practical deployment remains constrained by latency, privacy, and resource limitations, particularly under edge-computing conditions. This paper presents the design and demonstrator-based deployment of a hybrid edge-based action detection system that combines skeleton-based motion analysis with vision-language models for semantic scene interpretation. Skeleton-based processing enables continuous, privacy-aware monitoring with low computational overhead, while vision-language models provide contextual understanding and zero-shot reasoning capabilities for complex and previously unseen situations. Rather than proposing new recognition models, the contribution focuses on a system-level comparison of both paradigms under realistic edge constraints. The system is implemented on a GPU-enabled edge device and evaluated with respect to latency, resource usage, and operational trade-offs using a demonstrator-based setup. The results highlight the complementary strengths and limitations of motioncentric and semantic approaches and motivate a hybrid architecture that selectively augments fast skeletonbased detection with higher-level semantic reasoning. The presented system provides a practical foundation for privacy-aware, real-time video analysis in public safety applications.
