Search

Amidst the domestic labor shortage and worldwide pandemic in recent years, there has been an urgent need for a digital means that allows construction site workers, particularly site managers, to obtain information more efficiently in support of their daily managerial tasks. For workers who move around the site, traditional software applications that rely on a form-based interface and require multiple finger movements such as key hits and clicks can be inconvenient and reduce their willingness to use such applications. Conversational AI, also known as a chatbot, can improve the ease of use and usability of a system by providing an intuitive interface for user input. This study presents a demonstrative Natural Language Understanding (NLU) model and prototypes an AI-based chatbot for site managers to inquire about building component dimensions during their daily routines. Building Information Modeling (BIM) techniques are also applied to implement the answering module of the chatbot. The preliminary testing results show that the chatbot can successfully predict the intents and entities behind the inquiries raised by site managers with satisfactory accuracy for both intent prediction and the answer. These results provide site managers with alternative means to retrieve the information they need.

The indoor space model is the foundation of most indoor location-based services (LBS). A complete indoor space model includes floor-level paths and non-level paths. The latter includes passages connecting different floors or elevations such as stairs, elevators, escalators, and ramps. Most related studies have merely discussed the modeling and generation of floor-level paths, while those considering non-level paths usually simplify the formation and generation of non-level paths, especially stairs, which play an important role in emergency evacuation and response. Although the algorithm proposed by i-GIT approach, which considers both floor-level and non-level paths, can automatically generate paths of straight stairs, it is not applicable to the spiral stairs and winder stairs that are common in town houses and other public buildings. This study proposes a novel approach to generate high-accuracy stair paths that can support straight, spiral, and winder stairs. To implement and verify the proposed algorithm, 54 straight and spiral stairs provided by Autodesk Revit’s official website and three self-built winder stairs are used as test cases. The test results show that the algorithm can successfully produce the stair paths of most test cases (49/50), which comprehensively extends the applicability of the proposed algorithm.

Both humans and non-human animals exhibit sensitivity to the approximate number of items in a visual array, as indexed by their performance in numerosity discrimination tasks, and even neonates can detect changes in numerosity. These findings are often interpreted as evidence for an innate 'number sense'. However, recent simulation work has challenged this view by showing that human-like sensitivity to numerosity can emerge in deep neural networks that build an internal model of the sensory data. This emergentist perspective posits a central role for experience in shaping our number sense and might explain why numerical acuity progressively increases over the course of development. Here we substantiate this hypothesis by introducing a progressive unsupervised deep learning algorithm, which allows us to model the development of numerical acuity through experience. We also investigate how the statistical distribution of numerical and non-numerical features in natural environments affects the emergence of numerosity representations in the computational model. Our simulations show that deep networks can exhibit numerosity sensitivity prior to any training, as well as a progressive developmental refinement that is modulated by the statistical structure of the learning environment. To validate our simulations, we offer a refinement to the quantitative characterization of the developmental patterns observed in human children. Overall, our findings suggest that it may not be necessary to assume that animals are endowed with a dedicated system for processing numerosity, since domain-general learning mechanisms can capture key characteristics others have attributed to an evolutionarily specialized number system.

Abstract This study evaluates the performance of ChatGPT variants, GPT-3.5 and GPT-4, both with and without prompt engineering, against solely student work and a mixed category containing both student and GPT-4 contributions in university-level physics coding assignments using the Python language. Comparing 50 student submissions to 50 AI-generated submissions across different categories, and marked blindly by three independent markers, we amassed $$n = 300$$ n = 300 data points. Students averaged 91.9% (SE:0.4), surpassing the highest performing AI submission category, GPT-4 with prompt engineering, which scored 81.1% (SE:0.8)—a statistically significant difference (p = $$2.482 \times 10^{-10}$$ 2.482 × 10 - 10 ). Prompt engineering significantly improved scores for both GPT-4 (p = $$1.661 \times 10^{-4}$$ 1.661 × 10 - 4 ) and GPT-3.5 (p = $$4.967 \times 10^{-9}$$ 4.967 × 10 - 9 ). Additionally, the blinded markers were tasked with guessing the authorship of the submissions on a four-point Likert scale from ‘Definitely AI’ to ‘Definitely Human’. They accurately identified the authorship, with 92.1% of the work categorized as ‘Definitely Human’ being human-authored. Simplifying this to a binary ‘AI’ or ‘Human’ categorization resulted in an average accuracy rate of 85.3%. These findings suggest that while AI-generated work closely approaches the quality of university students’ work, it often remains detectable by human evaluators.

Summary: Background: Genome sequencing (GS) has revolutionised the diagnostic odyssey of patients with rare genetic diseases (RDs) and accelerated large-scale genome projects globally. However, the impact of GS on patients with RDs is yet to be investigated among genome projects in Asia. The Hong Kong Genome Project (HKGP) was implemented to benefit patients and families with RDs in Hong Kong, and to increase the inclusiveness of Chinese genomic data. This study evaluated the impact of short read GS (srGS), complemented by long read GS (lrGS) in a subset, on individuals recruited in the pilot phase of the HKGP. Methods: GS was performed on a prospective cohort of patients with suspected genetic disease recruited by territory-wide referrals to the HKGP. All participants received srGS, while lrGS was applied to a subset to resolve technically challenging regions unclear from srGS and provide phasing information for potential compound heterozygous variants. A phenotypic-driven diagnostic workflow was implemented to filter and prioritise rare and likely disease-causing variants. The primary outcome was diagnostic yield. The impact on the diagnostic odyssey and clinical management was also assessed. Findings: A total of 1264 individuals from 520 families with a broad spectrum of RDs were recruited, with 94% of probands being Chinese. srGS was performed for all individuals and lrGS was performed in 21 individuals. The use of srGS achieved a molecular diagnosis in 24% (125/520) of probands, and an additional 4% (21/520) with the assistance from lrGS. Approximately one-third of the identified diagnostic variants being novel. Diagnostic yield was found to be significantly higher among adult probands compared to paediatric probands (32% vs 24%; p = 0.025). The diagnostic yield was significantly higher in probands without prior genetic testing (37%; n = 185) compared to those previously tested, including exome and genome sequencing (23%; n = 335) (p = 0.001). GS ended diagnostic odysseys with an average length of 15 years (0.5–59), and potentially impacted clinical management in 77% (113/146) of diagnosed probands. Interpretation: This population-based genome project shed light on the consideration of integrating srGS and lrGS in clinical workflows for RDs. The identification of unique and prevalent variants from Southeast Asia increased the inclusiveness of Chinese genomic data, contributing to greater representation and genomic diversity. Funding: The HKGP is a publicly funded genome sequencing initiative commissioned by the Health Bureau of the HKSAR Government.

El presente artículo tiene como finalidad analizar las oportunidades de la fintech frente a la crisis económica generada por el COVID-19, la cual ha tenido un gran impacto en las empresas y la vida de las personas que se han visto obligadas a reinventar su trabajo, hábitos de consumo e interacción con el entorno para sostenerse en un mercado cada vez más competitivo que afecta su estilo de vida. Es fundamental tener en cuenta que el sistema tradicional y el gobierno han realizado esfuerzos por mantener la estabilidad económica del país, pero los resultados no han sido muy satisfactorios. Por ello, se lleva a cabo una investigación exploratoria de tipo triangular desde lo cualitativo y cuantitativo, mediante la aplicación de instrumentos de recolección de información, que identifica factores financieros, organizacionales y estratégicos de las pymes en la ciudad de Bogotá, que reportaron sus estados financieros con corte a diciembre 2021 a la Superintendencia de Sociedades. Por último, se evidencia la oportunidad que tienen las fintech de hacer innovación tecnológica disruptiva para el desarrollo de productos financieros por medio de la conectividad desde dispositivos móviles de uso personal que permitan el monitoreo, el comercio electrónico, sistemas de transacciones, de financiación, modelos de toma de decisiones y optimización de procesos financieros a un bajo costo, lo que posibilita abrir un abanico de múltiples oportunidades para las pymes que buscan mayor participación y competitividad, donde puedan ofertar sus productos y servicios de forma eficiente, ágil, cómoda y confiable.