12560 research outputs found
Sort by
Injecting commonsense knowledge into prompt learning for zero-shot text classification
The combination of pre-training and fine-tuning has become a default solution to Natural Language Processing (NLP) tasks. The emergence of prompt learning breaks such routine, especially in the scenarios of low data resources. Insufficient labelled data or even unseen classes are frequent problems in text classification, equipping Pre-trained Language Models (PLMs) with task-specific prompts helps get rid of the dilemma. However, general PLMs are barely provided with commonsense knowledge. In this work, we propose a KG-driven verbalizer that leverages commonsense Knowledge Graph (KG) to map label words with predefined classes. Specifically, we transform the mapping relationships into semantic relevance in the commonsense-injected embedding space. For zero-shot text classification task, experimental results exhibit the effectiveness of our KG-driven verbalizer on a Twitter dataset for natural disasters (i.e. HumAID) compared with other baselines
Professional development and recognition in LD
This chapter addresses professional development and recognition opportunities within the Learning Development field. It initially provides a chronology of how bespoke professional development for Learning Developers expanded during the 2010s and details where such opportunities can be accessed. This is followed by an overview of the ALDinHE professional recognition scheme along with guidance for potential applicants. The chapter concludes with consideration of how Learning Developers can acquire the experience and expertise necessary to apply for wider sector recognition. Collectively, the chapter is intended to inform and support a Learning Developer continued professional development planning and career progression
The English street gang and government policy
John Pitts outlines governmental responses to ‘gang’ and serious youth violence in the twenty-first century. He observes that although gang-related knife and gun crime among children and young people of African Caribbean descent was rising rapidly in London from 2004, it was the murder of Rhys Jones, an 11 year-old White boy, in Liverpool, in August 2007 that caused Prime Minister Gordon Brown to launch the Tackling Gangs Action Programme (TGAP). Later, the 2011 riots triggered the Ending Gang and Youth Violence programme of 2012, initially targeting 30 areas and extended to a further 22 in 2016. Evaluations of the programme were equivocal but by now the government’s focus had shifted to child sexual and criminal exploitation and County Lines drug dealing, withdrawing funds from local authorities and handing responsibility to the National Crime Agency. Nonetheless County Lines proliferated and knife crime soared, leading PM Theresa May to convene a Serious Youth Violence Summit in February 2019. Maintaining that we could not ‘arrest our way out of the problem’ May placed a requirement on health and welfare agencies to prioritise youth violence prevention. However, July 2019 saw the election of Boris Johnson and Home Secretary Priti Patel who promised more stop and searches, more police and more prison places
Effectiveness of postnatal maternal or caregiver interventions on outcomes among infants under six months with growth faltering
* Commissioned report submitted to the World Health Organisation. Reports 57-60 has been used to inform WHO Guideline on the prevention and management of wasting and nutritional oedema (acute malnutrition): Management section | Global Nutrition Cluster: Technical Allianc
Cloaking class: making the working class visible
Since the 1970s, wealth inequality in the UK has been rising year on year with the only exception being the early 2000s when the Blair Labour government made a few tweaks through their tax credit policy, slowing down the rapid widening of the wealth gap for a few short years. The Conservative/Liberal Democrat coalition government's austerity measures in 2010 speeded the process up to such an extent that by 2014 the gap between the wealthy and the poor had never been greater (Hills, 2014). Consequently, it should come as no surprise that in the country where the Industrial Revolution began – and, as E.P. Thompson chronicled, the English working class were made – the vertical class system that capitalism needs to function is in great shape. There has been a redistribution of wealth from the poorest upwards since 2010 (ONS, 2021). The divisions between the rich and the poor both in the UK and in Europe have never been greater in modern times, and there is now a full body of work from academics studying economics, sociology, anthropology and urbanism, focusing on the elites and the evergrowing wealth gaps particularly in the UK and within Europe (Savage, 2010, 2015, 2021; Picketty, 2014; Atkinson et al, 2017). However, the narrative of a solution to the widening wealth gap has focused somewhat on social mobility and meritocracy. To an extent, these concepts acknowledge the structured nature of inequality within capitalist societies, but for the most part social mobility is still seen as an individual act of pulling yourself up by the bootstraps. The social mobility narratives favoured by governments and administrations focus on shaping and introducing specific policies targeted at the bottom of society with the intention of creating ‘fairer opportunities’, with the hope that this will change the fates of some working-class people who are ‘hardworking’, ‘naturally clever’, and who accept without complaint the dominant narratives and culture (Friedman and Laurison, 2019; Social Mobility Commission, 2021). However, these policies focusing on social mobility are not intended to change the system, nor to tackle the unfair advantages the system awards to those in the middle or at the top
Building resilience in social work: a multi-level approach
Social care work is rewarding, but complex and emotionally demanding. The intrinsic pressures of the role combined with the broader context of contemporary social work presents many challenges and practitioners are vulnerable to work-related stress and burnout. This has negative implications not only for their own wellbeing, but also their ability to work effectively with service users. In this chapter, we focus on the need for a systemic, multi-level approach to building resilience in social work organizations. Firstly, we provide an overview of literature that has examined the wellbeing of social workers and the key risk factors and then highlight the potential for resilience to help them manage the emotional hazards of practice and remain healthy and effective. While resilience is typically characterised by the ability of individuals to learn and adapt to adversity, it is only sustainable when underpinned by supportive organizational policies and practices. We maintain that what is considered "resilience" is shaped by the context within which people operate-the demands they experience, the qualities they require do their job effectively, and the features of their working environment. Drawing on research conducted by ourselves and others, we set out a systemic approach that identifies the meaning of resilience in social work settings and the individual and organizational factors that underpin it. We also identify actions that can be taken at individual, team and organizational levels to foster a workplace climate that builds the capacity for resilience among social workers to support their wellbeing and optimum practice
Sounding others’ sensations in interaction
This study investigates the practice of “sounding for others,” wherein one person vocalizes to enact someone else’s putatively ongoing bodily sensation. We argue that it constitutes a collaborative way of performing sensorial experiences. Examples include producing cries with others’ strain or pain and parents sounding an mmm of gustatory pleasure on their infant’s behalf. Vocal sounds, their loudness, and duration are specifically deployed for instructing bodily experiences during novices’ real-time performance of various activities, such as tasting food for the first time or straining during a Pilates exercise. Vocalizations that are indexically tied to the body provide immediate displays of understanding and empathy that may be explicated further through lexicon. The existence of this practice challenges the conceptualization of communication as a transfer of information from an individual agent–even regarding assumedly individual body sensations–instead providing evidence of the joint nature of action and supporting dialogic theories of communication, including when language-marginal vocalizations are used
When a habit meets a habit at the city dump: persistence and adaptability of habit in Única mirando al mar (2010)
Fernando Contreras Castro’s novel Única mirando al mar (1993 and 2010) is a bitter, scathing social critique of Costa Rica’s government which occasionally remembers its duty to its tax payers and environment. Re-written in 2010 after the closure of Río Azul, a landfill near San José, it explores the way this action affected the population of the landfill, locally known as the buzos (scuba divers). The critiques of the novel address primarily the buzos’ resilience and the complexity of the relationship between the buzos’ society and the rest of the country. This study will consider the buzos’ society as a social order based on habit and trace its adaptability and persistence in face of multiple changes to the buzos’ life. To this end, I will use the theory of posthegemony with a particular focus on Pierre Bourdieu’s concepts of habit and habitus to explore the adaptability and persistence of the buzos’ habitus as reflected in habitual practices, rituals, and ritual language. I will analyse the 2010 version of the novel, published after the closure of the landfill
Widening the care gap? an international comparison of care-leaving in the time of COVID-19
Following COVID-19, young people who transitioned to adulthood from different types of alternative care (care-leavers) experienced an exacerbation of the challenges they had before the pandemic. The purpose of this international survey was to explore the range of policy and service responses that have or have not been implemented around the world to support care-leavers during COVID-19. Responses were collected from care-leaving researchers from 19 countries towards the end of 2020. Half of the participating countries reported that the state had issued directives about measures that should be taken to support care-leavers following COVID-19 outbreak, but only three reported actual changes in legislation. Additionally, NGOs in various countries took steps to guide and support care-leavers, while two thirds reported on special initiatives that were mounted. The most common change in practices during COVID-19 was the postponement of exits from care, and the second was an increase in contact from workers. These findings are critically discussed in relation to the impact of policy changes on an already vulnerable group. In particular, we indicate that there appears to be a widening care-gap: some countries with stronger leaving-care legal and policy frameworks pre-COVID-19 were more inclined to introduce additional supportive measures during the pandemic, whereas some with under-developed services tended not to increase support. By contrast, other countries used this crisis to develop services that were not available before. The creativity and flexibility in the services provided during the COVID-19 outbreak are required on an ongoing basis and thus should be implemented overall
KEMIM: knowledge-enhanced user multi-interest modeling for recommender systems
Researchers typically leverage side information, such as social networks or the knowledge graph, to overcome the sparsity and cold start problem in collaborative filtering. To tackle the limitations of existing user interest modeling, we propose a knowledge-enhanced user multi-interest modeling for recommender systems (KEMIM). First, we utilize the user-item historical interaction as the knowledge graph’s head entity to create a user’s explicit interests and leverage the relationship path to expand the user’s potential interests through connections in the knowledge graph. Second, considering the diversity of a user’s interests, we adopt an attention mechanism to learn the user’s attention to each historical interaction and each potential interest. Third, we combine the user’s attribute features with interests to solve the cold start problem effectively. With the knowledge graph’s structural data, KEMIM could describe the features of users at a fine granularity and provide explainable recommendation results to users. In this study, we conduct an in-depth empirical evaluation across three open datasets for two different recommendation tasks: Click-Through rate (CTR) prediction and Top-K recommendation. The experimental findings demonstrate that KEMIM outperforms several state-of-the-art baselines