Appl. Syst. Innov., Volume 7, Issue 1 (February 2024) – 17 articles

This paper addresses the challenges associated with the centralized storage of educational materials in the context of a fragmented and disparate database. In response to the increasing demands of modern education, efficient and accessible retrieval of materials for educators and students is essential. This paper presents a hybrid model based on the transformer framework and utilizing an API for an existing large language model (LLM)/chatbot. This integration ensures precise responses drawn from a comprehensive educational materials database. The model architecture uses mathematically defined algorithms for precise functions that enable deep text processing through advanced word embedding methods. This approach improves accuracy in natural language processing and ensures both high efficiency and adaptability. Therefore, this paper not only provides a technical solution to a prevalent problem but also highlights the potential for the continued development and integration of emerging technologies in education. The aim is to create a more efficient, transparent, and accessible educational environment. The importance of this research lies in its ability to streamline material access, benefiting the global scientific community and contributing to the continuous advancement of educational technology. Full article

Traditional methods of supervision in the construction industry are time-consuming and costly, requiring significant investments in skilled labor. However, with advancements in artificial intelligence, computer vision, and deep learning, these methods can now be automated, resulting in time and cost savings, as well as improvements in product quality. This research focuses on the application of computer vision approaches to monitor the quality of welding in prefabricated steel elements. A high-performance network was designed, consisting of a video capturing station, a customized classifier based on a YOLOv4 detector and an IoU tracker, and a user interface software for any interaction with quality control workers. The network demonstrated over 98% accuracy in identifying steel connection types and detecting missed welds on the assembly line in real-time. Extensive validation was conducted using a large dataset from a real production environment. The proposed framework aims to reduce rework, minimize hazards, and enhance product quality. This research contributes to the automation of quality control processes in the construction industry. Full article

Soft pneumatic grippers can grasp soft or irregularly shaped objects, indicating potential applications in industry, agriculture, and healthcare. However, soft grippers rarely carry heavy and dense objects due to the intrinsic low modulus of soft materials in nature. This paper designed a soft robotic gripper with rigid supports to enhance lifting force by 150 ± 20% in comparison with that of the same gripper without supports, which successfully lifted a metallic wrench (672 g). The soft gripper also achieves excellent adaptivity for irregularly shaped objects. The design, fabrication, and performance of soft grippers with rigid supports are discussed in this paper. Full article

In an era defined by technological disruption, the integration of artificial intelligence (AI) into business processes is both strategic and challenging. As AI continues to disrupt and reshape industries and revolutionize business processes, organizations must take proactive steps to assess their readiness and capabilities to effectively leverage AI technologies. This research focuses on the assessment elements required to evaluate an organization’s current state in preparation for AI-based digital transformation. This research is based on a literature review and practical insights derived from extensive experience in industrial system engineering. This paper outlines the key assessment elements that organizations should consider to ensure successful and sustainable AI-based digital transformation. This emphasizes the need for a comprehensive approach to assess the organization’s data infrastructure, governance practices, and existing AI capabilities. Furthermore, the research work focuses on the evaluation of AI talent and skills within the organization, considering the significance of fostering an innovative culture and addressing change management challenges. The results of this study provide organizations with elements to assess their current state for AI-based digital transformation. By adopting and implementing the proposed guidelines, organizations can gain a holistic perspective of their current standing, identify strategic opportunities for AI integration, mitigate potential risks, and strategize a successful path forwards in the evolving landscape of AI-driven digital transformation. Full article

Artificial intelligence (AI) has become popular worldwide after technological breakthroughs in the early 2010s. Accordingly, many organizations and individuals have been using AI for various applications. Previous research has been dominated by case studies regarding the industrial use of AI, although how time-series changes affect users’ perceptions has not been clarified yet. This study analyzes time-series changes in AI perceptions through text mining from nonfinancial information obtained from Japanese firms’ disclosures. The main findings of this study are as follows: first, perceptions of AI vary across industries; second, the business sector has progressed through the stages of recognition, investment, strategization, commercialization, and monetization. This transition is concurrent with each category’s evolving interpretation of the innovator theory proposed by Rogers (2003), to some extent. Third, it took approximately a decade from the breakthrough technology to the monetization by Japanese firms. Our findings underline the importance of speeding up the organizational process through intervention and contribution to the areas regarding “diffusion of innovation” and perceptual characteristics. Full article

Academic advising is often pivotal in shaping students’ educational experiences and choices. This study leverages natural language processing to quantitatively evaluate reviews of academic advisors, aiming to provide actionable insights on key feedback phrases and demographic factors for enhancing advising services. This analysis encompassed a comprehensive evaluation of 1151 reviews of undergraduate students for academic advisors, which were collected within a European University alliance consisting of five universities, offering a diverse pool of feedback from a wide range of academic interactions. Employing sentiment analysis powered by artificial intelligence, we computed compound sentiment scores for each academic advisor’s reviews. Subsequently, statistical analyses were conducted to provide insights into how demographic factors may or may not influence students’ sentiment and evaluations of academic advisory services. The results indicated that advisor’s gender had no substantial influence on the sentiment of the reviews. On the contrary, the academic advisors’ age showed a notable impact, with younger advisors surprisingly receiving more favorable evaluations. Word frequency analyses, both for positive and negative expressions, were also performed to contextualize the language used in describing academic advisors. The prevalent word combinations in reviews of highly rated academic advisors emphasized attributes like empathy, approachability, and effectiveness in guiding students towards achieving their academic goals. Conversely, advisors with less favorable reviews were often perceived as inadequate in addressing students’ concerns related to their academic journey, revealing persistent challenges in the student–advisor interaction that impacted their evaluation. This analysis of academic advisor reviews contributes to the body of literature by highlighting the significance of managing student expectations and enhancing advisor skills and qualities to foster positive interactions and academic success. Full article

Quality assessment in industrial applications is often carried out through visual inspection, usually performed or supported by human domain experts. However, the manual visual inspection of processes and products is error-prone and expensive. It is therefore not surprising that the automation of visual inspection in manufacturing and maintenance is heavily researched and discussed. The use of artificial intelligence as an approach to visual inspection in industrial applications has been considered for decades. Recent successes, driven by advances in deep learning, present a possible paradigm shift and have the potential to facilitate automated visual inspection, even under complex environmental conditions. For this reason, we explore the question of to what extent deep learning is already being used in the field of automated visual inspection and which potential improvements to the state of the art could be realized utilizing concepts from academic research. By conducting an extensive review of the openly accessible literature, we provide an overview of proposed and in-use deep-learning models presented in recent years. Our survey consists of 196 open-access publications, of which 31.7% are manufacturing use cases and 68.3% are maintenance use cases. Furthermore, the survey also shows that the majority of the models currently in use are based on convolutional neural networks, the current de facto standard for image classification, object recognition, or object segmentation tasks. Nevertheless, we see the emergence of vision transformer models that seem to outperform convolutional neural networks but require more resources, which also opens up new research opportunities for the future. Another finding is that in 97% of the publications, the authors use supervised learning techniques to train their models. However, with the median dataset size consisting of 2500 samples, deep-learning models cannot be trained from scratch, so it would be beneficial to use other training paradigms, such as self-supervised learning. In addition, we identified a gap of approximately three years between approaches from deep-learning-based computer vision being published and their introduction in industrial visual inspection applications. Based on our findings, we additionally discuss potential future developments in the area of automated visual inspection. Full article

In computer vision and image processing, the shift from traditional cameras to emerging sensing tools, such as gesture recognition and object detection, addresses privacy concerns. This study navigates the Integrated Sensing and Communication (ISAC) era, using millimeter-wave signals as radar via a Convolutional Neural Network (CNN) model for event sensing. Our focus is on leveraging deep learning to detect security-critical gestures, converting millimeter-wave parameters into point cloud images, and enhancing recognition accuracy. CNNs present complexity challenges in deep learning. To address this, we developed flexible quantization methods, simplifying You Only Look Once (YOLO)-v4 operations with an 8-bit fixed-point number representation. Cross-simulation validation showed that CPU-based quantization improves speed by 300% with minimal accuracy loss, even doubling the YOLO-tiny model’s speed in a GPU environment. We established a Raspberry Pi 4-based system, combining simplified deep learning with Message Queuing Telemetry Transport (MQTT) Internet of Things (IoT) technology for nursing care. Our quantification method significantly boosted identification speed by nearly 2.9 times, enabling millimeter-wave sensing in embedded systems. Additionally, we implemented hardware-based quantization, directly quantifying data from images or weight files, leading to circuit synthesis and chip design. This work integrates AI with mmWave sensors in the domain of nursing security and hardware implementation to enhance recognition accuracy and computational efficiency. Employing millimeter-wave radar in medical institutions or homes offers a strong solution to privacy concerns compared to conventional cameras that capture and analyze the appearance of patients or residents. Full article

Plantar pressure plays a crucial role in the pathogenesis of foot ulcers among patients with diabetes and peripheral polyneuropathy. Pressure relief is a key requirement for both the prevention and treatment of plantar ulcers. Conventional medical practice to enable such an action is usually realized by means of dedicated insoles and special footwear. Another technique for foot pressure offloading (not in medical practice) can be achieved by sensing/estimating the current state (pressure) and, accordingly, enabling a pressure release mechanism once a defined threshold is reached. Though these mechanisms can make plantar pressure monitoring and release possible, overall, they make shoes bulkier, power-dependent, and expensive. In this work, we present a passive and self-offloading alternative to keep plantar pressure within a defined safe limit. Our approach is based on the use of a permanent magnet, taking advantage of its non-linear field reduction with distance. The proposed solution is free from electronics and is a low-cost alternative for smart shoe development. The overall size of the device is 13 mm in diameter and 30 mm in height. The device allows more than 20-times the tunability of the threshold pressure limit, which makes it possible to pre-set the limit as low as 38 kPa and as high as 778 kPa, leading to tunability within a wide range. Being a passive, reliable, and low-cost alternative, the proposed solution could be useful in smart shoe development to prevent foot ulcer development. The proposed device provides an alternative for offloading plantar pressure that is free from the power feeding requirement. The presented study provides preliminary results for the development of a complete offloading shoe that could be useful for the prevention/care of foot ulcers among diabetic patients. Full article

The task of named entity recognition (NER) is to identify entities in the text and predict their categories. In real-life scenarios, the context of the text is often complex, and there may exist nested entities within an entity. This kind of entity is called a nested entity, and the task of recognizing entities with nested structures is referred to as nested named entity recognition. Most existing NER models can only handle flat entities, and there has been limited research progress in Chinese nested named entity recognition, resulting in relatively few models in this direction. General NER models have limited semantic extraction capabilities and cannot capture deep semantic information between nested entities in the text. To address these issues, this paper proposes a model that uses the GlobalPointer module to identify nested entities in the text and constructs the IDCNNLR semantic extraction module to extract deep semantic information. Furthermore, multiple-head self-attention mechanisms are incorporated into the model at multiple positions to achieve data denoising, enhancing the quality of semantic features. The proposed model considers each possible entity boundary through the GlobalPointer module, and the IDCNNLR semantic extraction module and multi-position attention mechanism are introduced to enhance the model’s semantic extraction capability. Experimental results demonstrate that the proposed model achieves F1 scores of 69.617% and 79.285% on the CMeEE Chinese nested entity recognition dataset and CLUENER2020 Chinese fine-grained entity recognition dataset, respectively. The model exhibits improvement compared to baseline models, and each innovation point shows effective performance enhancement in ablative experiments. Full article

Accreditation bodies call for curriculum development processes that are open to all stakeholders, reflecting viewpoints of students, industry, university faculty, and society. However, communication difficulties between faculty and non-faculty groups leave an immense collaboration potential unexplored. Using the classification of learning objectives, natural language processing, and data visualization, this paper presents a quantitative method that delivers program plan representations that are universal, self-explanatory, and empowering; promoting stronger links between program courses and curriculum development open to all stakeholders. A simple example shows how the method contributes to representative program planning experiences and a case study is used to confirm the method’s accuracy and utility. Full article

Over the past few decades, the education sector has achieved impressive advancements by incorporating Artificial Intelligence (AI) into the educational environment. Nevertheless, specific educational processes, particularly educational counseling, still depend on traditional procedures. The current method of conducting group sessions between counselors and students does not offer personalized assistance or individual attention, which can cause stress to students and make it difficult for them to make informed decisions about their coursework and career path. This paper proposes a counseling solution designed to aid high school seniors in selecting appropriate academic paths at the tertiary level. The system utilizes a predictive model that considers academic history and student preferences to determine students’ likelihood of admission to their chosen university and recommends similar alternative universities to provide more opportunities. We developed the model based on data from 500 graduates from 12 public high schools in Morocco, as well as eligibility criteria from 31 institutions and colleges. The counseling system comprises two modules: a recommendation module that uses popularity-based and content-based recommendations and a prediction module that calculates the likelihood of admission using the Huber Regressor model. This model outperformed 13 other machine learning modules, with a low MSE of 0.0017, RMSE of 0.0422, and the highest R-squared value of 0.9306. Finally, the system is accessible through a user-friendly web interface. Full article

This article is devoted to the method of numerical modelling of aerodynamics when the air flows around fins of a special design, which is implemented in SolidWorks Flow Simulation. The study was carried out for three types of rib orientation, and the aerodynamic drag coefficients were determined for different values of the Reynolds number. It was confirmed that the drag coefficient values depend significantly on the flow regime. The lowest value of the drag coefficient is observed when the fins are oriented from a larger diameter to a smaller one. In the laminar regime (Re < 2300), the average value of C_X = 1.04, in the transitional regime (2300 < Re < 10,000), C_X = 0.74, and in the turbulent regime (Re > 10,000), C_X = 0.22. Characteristic for this case of orientation is a significant decrease in the drag coefficient during the transition from laminar to turbulent regime; the minimum is observed at the flow speed in the range between 2 and 3 m/s. Full article

The mathematical modeling and the associated numerical simulation of the light–matter interaction (LMI) process are well-known to be quite complicated, particularly for media where several electronic transitions take place under electromagnetic excitation. As a result, numerical simulations of typical LMI processes usually require a high computational cost due to the involvement of a large number of coupled differential equations modeling electron and photon behavior. In this paper, we model the general LMI process involving an electromagnetic interaction medium and optical (light) excitation in one dimension (1D) via the use of a dynamic deep learning algorithm where the neural network coefficients can precisely adapt themselves based on the past values of the coefficients of adjacent layers even under the availability of very limited data. Due to the high computational cost of LMI simulations, simulation data are usually only available for short durations. Our aim here is to implement an adaptive deep learning-based model of the LMI process in 1D based on available temporal data so that the electromagnetic features of LMI simulations can be quickly decrypted by the evolving network coefficients, facilitating self-learning. This enables accurate prediction and acceleration of LMI simulations that can run for much longer durations via the reduction in the cost of computation through the elimination of the requirement for the simultaneous computation and discretization of a large set of coupled differential equations at each simulation step. Our analyses show that the LMI process can be efficiently decrypted using dynamic deep learning with less than 1% relative error (RE), enabling the extension of LMI simulations using simple artificial neural networks. Full article

Reducing CO₂ emissions is currently a key policy in most developed countries. In this article, we evaluate whether smart traffic lights can have a relevant role in reducing CO₂ emissions in small cities, considering their specific traffic profiles. The research method is a quantitative modelling approach tested by computational simulation. We propose a novel microscopic traffic simulation framework, designed to simulate realistic vehicle kinematics and driver behaviour, and accurately estimate CO₂ emissions. We also propose and evaluate a routing algorithm for smart traffic lights, specially designed to optimize CO₂ emissions at intersections. The simulations reveal that deploying smart traffic lights at a single intersection can reduce CO₂ emissions by 32% to 40% in the vicinity of the intersection, depending on the traffic density. The simulations show other advantages for drivers: an increase in average speed of 60% to 101% and a reduction in waiting time of 53% to 95%. These findings can be useful for city-level decision makers who wish to adopt smart technologies to improve traffic flows and reduce CO₂ emissions. This work also demonstrates that the simulator can play an important role as a tool to study the impact of smart traffic lights and foster the improvement in smart routing algorithms to reduce CO₂ emissions. Full article

The research question this study poses is how to measure the efficiency of project management activities. The purpose of this article is to quantify the efficiency of the execution of a project portfolio managed by a project management office (PMO) structure. The research subject is a PMO operating within a petrochemical manufacturing company in southern Brazil. The research method is quantitative modeling. The study employed data envelopment analysis (DEA) to calculate the relative efficiencies of projects in three classes according to complexity over a period of four years. Each project is a decision-making unit (DMU), as required by the DEA procedure. One novelty is the calculation of cost- and time-weighted efficiency values, which slightly differ from the average. The main results indicate that the average efficiency for classes of projects roughly stands between 40 and 80%. The results also indicate a learning process guided by the PMO, as the average efficiency increased over three years in two classes of projects, according to the prioritization imposed by the office. The study also pointed out that the most influential variables in determining project efficiency are accuracy in meeting deadlines and the time planned for completion. The most important implication is that, from now on, the company has a theoretical foundation to justify focusing further efforts on reducing and controlling time to completion, not only cost and scope conformity, to increase overall project efficiency. Future research should prioritize investigating management techniques that increase the likelihood of completing projects within their deadlines. Full article

Artificial vision system applications have generated significant interest as they allow information to be obtained through one or several of the cameras that can be found in daily life in many places, such as parks, avenues, squares, houses, etc. When the aim is to obtain information from large areas, it can become complicated if it is necessary to track an object of interest, such as people or vehicles, due to the vision space that a single camera can cover; this opens the way to distributed zone monitoring systems made up of a set of cameras that aim to cover a larger area. Distributed zone monitoring systems add great versatility, becoming more complex in terms of the complexity of information analysis, communication, interoperability, and heterogeneity in the interpretation of information. In the literature, the development of distributed schemes has focused on representing data communication and sharing challenges. Currently, there are no specific criteria for information exchange and analysis in a distributed system; hence, different models and architectures have been proposed. In this work, the authors present a framework to provide homogeneity in a distributed monitoring system. The information is obtained from different cameras, where a global reference system is defined for generated trajectories, which are mapped independently of the model used to obtain the dynamics of the movement of people within the vision area of a distributed system, thus allowing for its use in works where there is a large amount of information from heterogeneous sources. Furthermore, we propose a novel similarity metric that allows for information queries from heterogeneous sources. Finally, to evaluate the proposed performance, the authors developed several distributed query applications in an augmented reality system based on realistic environments and historical data retrieval using a client–server model. Full article