Social welfare in the light of topic modelling

With an increased focus on social well-being in response to a burgeoning global economy exposing the weaknesses of social welfare policies, research output in the field has grown exponentially. Keeping track of the evolving research themes proves difficult due to the steady rise in the number of studies published in the interdisciplinary field of social welfare. Therefore, researchers need a comprehensive overview to confirm the current shape of the field based on the published research. Using a latent Dirichlet allocation algorithm as a topic modelling technique, this study identified 12 prominent themes from more than 10,000 research outputs on social welfare published from 2000 to 2020 in Scopus-indexed journals. Such an exploratory text-mining approach to literature review provides broad insights into the diversity of research and may serve as a foundation for further in-depth studies. Identifying these 12 thematic areas and their sub-themes allows us to articulate the complexity and diversity of social welfare issues, which go far beyond the field of well-established welfare economics or social work. The study shows that the topic of ‘social welfare’ has not only evolved over time but has significantly broadened its meaning. It can no longer be solely synonymous with institutional social security. We contend that research in this area needs to take into account a broader and more systematic range of determinants constituting the dynamic character of social welfare.     

基于机器学习的小微企业信贷供给智能决策系统研究——以“银税互动”政策的共享纳税数据为例

随着金融科技的快速发展,如何融合现代信贷技术有效缓解小微企业的融资困境是当前金融市场亟待解决的关键问题。以“银税互动”政策作为切入点,深入挖掘小微企业纳税数据的信用特征,结合随机森林方法和深度神经网络方法对小微企业的信用违约概率和信贷风险等级进行评估,综合考虑银行的经营利润以及风险成本实现不同小微企业信贷利率和信贷额度的优化,以此构建小微企业信贷供给智能决策系统。研究结果表明,纳税数据能够准确识别小微企业的信贷风险,从而为银行信贷供给提供决策依据,并且纳税数据质量较高的小微企业无论是信贷额度还是信贷利率都能够获得银行层面更加有利的信贷支持。     

Adaptive designs for best treatment identification with top-two Thompson sampling and acceleration

We consider outcome adaptive phase II or phase II/III trials to identify the best treatment for further development. Different from many other multi-arm multi-stage designs, we borrow approaches for the best arm identification in multi-armed bandit (MAB) approaches developed for machine learning and adapt them for clinical trial purposes. The best arm identification in MAB focuses on the error rate of identification at the end of the trial, but we are also interested in the cumulative benefit of trial patients, for example, the frequency of patients treated with the best treatment. In particular, we consider Top-Two Thompson Sampling (TTTS) and propose an acceleration approach for better performance in drug development scenarios in which the sample size is much smaller than that considered in machine learning applications. We also propose a variant of TTTS (TTTS2) which is simpler, easier for implementation, and has comparable performance in small sample settings. An extensive simulation study was conducted to evaluate the performance of the proposed approach in multiple typical scenarios in drug development.