As industrial recommender systems enter a scaling-driven regime, Transformer architectures have become increasingly attractive for scaling models towards larger capacity and longer sequence. However, existing Transformer-based recommendation models remain structurally fragmented, where sequence modeling and feature interaction are implemented as separate modules with independent parameterization. Such designs introduce a fundamental co-scaling challenge, as model capacity must be suboptimally allocated between dense feature interaction and sequence modeling under a limited computational budget. In this work, we propose MixFormer, a unified Transformer-style architecture tailored for recommender systems, which jointly models sequential behaviors and feature interactions within a single backbone. Through a unified parameterization, MixFormer enables effective co-scaling across both dense capacity and sequence length, mitigating the trade-off observed in decoupled designs. Moreover, the integrated architecture facilitates deep interaction between sequential and non-sequential representations, allowing high-order feature semantics to directly inform sequence aggregation and enhancing overall expressiveness. To ensure industrial practicality, we further introduce a user-item decoupling strategy for efficiency optimizations that significantly reduce redundant computation and inference latency. Extensive experiments on large-scale industrial datasets demonstrate that MixFormer consistently exhibits superior accuracy and efficiency. Furthermore, large-scale online A/B tests on two production recommender systems, Douyin and Douyin Lite, show consistent improvements in user engagement metrics, including active days and in-app usage duration.
Yingyan Zeng, Ismini Lourentzou, Xinwei Deng
et al.
Artificial intelligence (AI) systems have been increasingly adopted in the Manufacturing Industrial Internet (MII). Investigating and enabling the AI resilience is very important to alleviate profound impact of AI system failures in manufacturing and Industrial Internet of Things (IIoT) operations, leading to critical decision making. However, there is a wide knowledge gap in defining the resilience of AI systems and analyzing potential root causes and corresponding mitigation strategies. In this work, we propose a novel framework for investigating the resilience of AI performance over time under hazard factors in data quality, AI pipelines, and the cyber-physical layer. The proposed method can facilitate effective diagnosis and mitigation strategies to recover AI performance based on a multimodal multi-head self latent attention model. The merits of the proposed method are elaborated using an MII testbed of connected Aerosol Jet Printing (AJP) machines, fog nodes, and Cloud with inference tasks via AI pipelines.
3D anomaly detection (3D-AD) plays a critical role in industrial manufacturing, particularly in ensuring the reliability and safety of core equipment components. Although existing 3D datasets like Real3D-AD and MVTec 3D-AD offer broad application support, they fall short in capturing the complexities and subtle defects found in real industrial environments. This limitation hampers precise anomaly detection research, especially for industrial equipment components (IEC) such as bearings, rings, and bolts. To address this challenge, we have developed a point cloud anomaly detection dataset (IEC3D-AD) specific to real industrial scenarios. This dataset is directly collected from actual production lines, ensuring high fidelity and relevance. Compared to existing datasets, IEC3D-AD features significantly improved point cloud resolution and defect annotation granularity, facilitating more demanding anomaly detection tasks. Furthermore, inspired by generative 2D-AD methods, we introduce a novel 3D-AD paradigm (GMANet) on IEC3D-AD. This paradigm generates synthetic point cloud samples based on geometric morphological analysis, then reduces the margin and increases the overlap between normal and abnormal point-level features through spatial discrepancy optimization. Extensive experiments demonstrate the effectiveness of our method on both IEC3D-AD and other datasets.
Recommender systems have generated tremendous value for both users and businesses, drawing significant attention from academia and industry alike. However, due to practical constraints, academic research remains largely confined to offline dataset optimizations, lacking access to real user data and large-scale recommendation platforms. This limitation reduces practical relevance, slows technological progress, and hampers a full understanding of the key challenges in recommender systems. In this survey, we provide a systematic review of industrial recommender systems and contrast them with their academic counterparts. We highlight key differences in data scale, real-time requirements, and evaluation methodologies, and we summarize major real-world recommendation scenarios along with their associated challenges. We then examine how industry practitioners address these challenges in Transaction-Oriented Recommender Systems and Content-Oriented Recommender Systems, a new classification grounded in item characteristics and recommendation objectives. Finally, we outline promising research directions, including the often-overlooked role of user decision-making, the integration of economic and psychological theories, and concrete suggestions for advancing academic research. Our goal is to enhance academia's understanding of practical recommender systems, bridge the growing development gap, and foster stronger collaboration between industry and academia.
Background and objective Critical infrastructures, including water, electricity, and wastewater, are highly interdependent. Disruption in one of these systems can have a cascade effect on other systems and disrupt the entire system. In this research, to increase the resilience of these infrastructures against disasters, we aimed to model the interdependencies between them and prioritize critical nodes.
Method Based on the data for the urban network of Sioux Falls in South Dakota reported in a previous study, and taking into account its critical infrastructures, we used the DEMATEL method and Root Assessment Method (RAM) to identify the critical points of these networks for strengthening them. The study network had 21 nodes and there were five main criteria of node capacity, supply node, transmission node, connection with other networks (such as communication) and repair cost, and three sub-criteria of electricity, water and wastewater systems. To prioritize the nodes, we first obtained the weights of each criterion using the DEMATEL method. Then, using the RAM, we prioritized critical nodes that needed to be strengthened interdependent critical infrastructures in the pre-disaster phase.
Results Using the DEMATEL method, the results showed that the node capacity criterion was the most important criterion in decision-making. Also, in most cases, the electricity sub-criterion had the highest local weight. Using the RAM, the identified critical nodes were the nodes number 5, 6, 13, and 15. These nodes are usually transmission nodes or major facilities. Parameters such as capacity, repair time, and reinforcement cost play an important role in determining the importance of nodes.
Conclusion The node capacity, as the most influential factor, plays a key role in managing and improving urban critical infrastructures. The identified critical points require more attention. Reinforcement of these nodes can significantly increase the resilience and performance of urban critical infrastructures. Also, special attention should be paid to the interdependencies between critical infrastructures and necessary measures should be taken to reduce the interdependency.
Risk in industry. Risk management, Industrial safety. Industrial accident prevention
On modern industrial assembly lines, many intelligent algorithms have been developed to replace or supervise workers. However, we found that there were bottlenecks in both training datasets and real-time performance when deploying algorithms on actual assembly line. Therefore, we developed a promising strategy for expanding industrial datasets, which utilized large models with strong generalization abilities to achieve efficient, high-quality, and large-scale dataset expansion, solving the problem of insufficient and low-quality industrial datasets. We also applied this strategy to video action recognition. We proposed a method of converting hand action recognition problems into hand skeletal trajectory classification problems, which solved the real-time performance problem of industrial algorithms. In the "hand movements during wire insertion" scenarios on the actual assembly line, the accuracy of hand action recognition reached 98.8\%. We conducted detailed experimental analysis to demonstrate the effectiveness and superiority of the method, and deployed the entire process on Midea's actual assembly line.
Industrial machine fault diagnosis is a critical component of operational efficiency and safety in manufacturing environments. Traditional methods rely heavily on expert knowledge and specific machine learning models, which can be limited in their adaptability and require extensive labeled data. This paper introduces a novel approach leveraging Large Language Models (LLMs), specifically through a structured multi-round prompting technique, to improve fault diagnosis accuracy. By dynamically crafting prompts, our method enhances the model's ability to synthesize information from diverse data sources, leading to improved contextual understanding and actionable recommendations. Experimental results demonstrate that our approach outperforms baseline models, achieving an accuracy of 91% in diagnosing various fault types. The findings underscore the potential of LLMs in revolutionizing industrial fault consultation practices, paving the way for more effective maintenance strategies in complex environments.
Alexandre Trilla, Ossee Yiboe, Nenad Mijatovic
et al.
This paper describes the development of a causal diagnosis approach for troubleshooting an industrial environment on the basis of the technical language expressed in Return on Experience records. The proposed method leverages the vectorized linguistic knowledge contained in the distributed representation of a Large Language Model, and the causal associations entailed by the embedded failure modes and mechanisms of the industrial assets. The paper presents the elementary but essential concepts of the solution, which is conceived as a causality-aware retrieval augmented generation system, and illustrates them experimentally on a real-world Predictive Maintenance setting. Finally, it discusses avenues of improvement for the maturity of the utilized causal technology to meet the robustness challenges of increasingly complex scenarios in the industry.
Zhipeng Ma, Bo Nørregaard Jørgensen, Michelle Levesque
et al.
Digitalization is challenging in heavy industrial sectors, and many pi-lot projects facing difficulties to be replicated and scaled. Case studies are strong pedagogical vehicles for learning and sharing experience & knowledge, but rarely available in the literature. Therefore, this paper conducts a survey to gather a diverse set of nine industry cases, which are subsequently subjected to analysis using the business model canvas (BMC). The cases are summarized and compared based on nine BMC components, and a Value of Business Model (VBM) evaluation index is proposed to assess the business potential of industrial digital solutions. The results show that the main partners are industry stakeholders, IT companies and academic institutes. Their key activities for digital solutions include big-data analysis, machine learning algorithms, digital twins, and internet of things developments. The value propositions of most cases are improving energy efficiency and enabling energy flexibility. Moreover, the technology readiness levels of six industrial digital solutions are under level 7, indicating that they need further validation in real-world environments. Building upon these insights, this paper proposes six recommendations for future industrial digital solution development: fostering cross-sector collaboration, prioritizing comprehensive testing and validation, extending value propositions, enhancing product adaptability, providing user-friendly platforms, and adopting transparent recommendations.
Andrey Solano, Arne Sieverling, Robert Gieselmann
et al.
We present Fast-dRRT*, a sampling-based multi-robot planner, for real-time industrial automation scenarios. Fast-dRRT* builds upon the discrete rapidly-exploring random tree (dRRT*) planner, and extends dRRT* by using pre-computed swept volumes for efficient collision detection, deadlock avoidance for partial multi-robot problems, and a simplified rewiring strategy. We evaluate Fast-dRRT* on five challenging multi-robot scenarios using two to four industrial robot arms from various manufacturers. The scenarios comprise situations involving deadlocks, narrow passages, and close proximity tasks. The results are compared against dRRT*, and show Fast-dRRT* to outperform dRRT* by up to 94% in terms of finding solutions within given time limits, while only sacrificing up to 35% on initial solution cost. Furthermore, Fast-dRRT* demonstrates resilience against noise in target configurations, and is able to solve challenging welding, and pick and place tasks with reduced computational time. This makes Fast-dRRT* a promising option for real-time motion planning in industrial automation.
Yuiki Iwayama, Yuki Shimba, Chandra Sekhar Viswanathan
et al.
Objectives: Specific health guidance (SHG) has served as a preventive intervention for metabolic syndrome in Japan since 2008. For SHG, health professionals guide diet and physical activity to achieve body weight (BW) and waist circumference (WC) reductions. Since 2013, SHG intervention using information and communication technology (ICT-based SHG) has also been available. Therefore, in this study, we examined the effects of ICT-based SHG, and identified factors associated with BW and WC reductions in response to this intervention. Methods: Our intervention was performed using a smartphone application with videophone guidance and message exchanges provided by health professionals. We analysed 1,994 participants. Primary outcomes included changes in BW and WC after versus before the intervention. We used multiple linear regression analyses to identify factors associated with reductions in BW and WC due to the intervention. Results: The mean ages were 49.3 (standard deviation [SD], 5.8) years for males and 50.5 (SD, 5.8) years for females. The mean BW change was −1.37 kg for both sexes. The mean WC changes were −1.05 for males and −2.05 cm for females. For males, baseline body mass index, pre-intervention action history, and the numbers of videophone communications and messages were significantly associated with larger changes in BW and WC. For females, no factors were significant for BW reduction, while baseline WC and pre-intervention action history were associated with WC reduction. Conclusions: ICT-based SHG reduces BW and WC. Videophone communication and messaging are associated with reductions in BW and WC in males. These results may help to improve the efficacy of ICT-based SHG.
Industrial safety. Industrial accident prevention, Medicine (General)
Danny Weyns, Ilias Gerostathopoulos, Nadeem Abbas
et al.
Computing systems form the backbone of many areas in our society, from manufacturing to traffic control, healthcare, and financial systems. When software plays a vital role in the design, construction, and operation, these systems are referred as software-intensive systems. Self-adaptation equips a software-intensive system with a feedback loop that either automates tasks that otherwise need to be performed by human operators or deals with uncertain conditions. Such feedback loops have found their way to a variety of practical applications; typical examples are an elastic cloud to adapt computing resources and automated server management to respond quickly to business needs. To gain insight into the motivations for applying self-adaptation in practice, the problems solved using self-adaptation and how these problems are solved, and the difficulties and risks that industry faces in adopting self-adaptation, we performed a large-scale survey. We received 184 valid responses from practitioners spread over 21 countries. Based on the analysis of the survey data, we provide an empirically grounded overview of state-of-the-practice in the application of self-adaptation. From that, we derive insights for researchers to check their current research with industrial needs, and for practitioners to compare their current practice in applying self-adaptation. These insights also provide opportunities for the application of self-adaptation in practice and pave the way for future industry-research collaborations.
Past research on software product lines has focused on the initial development of reusable assets and related challenges, such as cost estimation and implementation issues. Naturally, as software product lines are increasingly adopted throughout industry, their ongoing maintenance and evolution are getting more attention as well. However, it is not clear to what degree research is following this trend, and where the interests and demands of the industry lie. In this technical report, we provide a survey and comparison of selected publications on software product line maintenance and evolution at SPLC. In particular, we analyze and discuss similarities and differences of these papers with regard to their affiliation with industry and academia. From this, we infer directions for future research that pave the way for systematic and organized evolution of software product lines, from which industry may benefit as well.
Industrial Internet of Things is an ultra-large-scale system that is much more sophisticated and fragile than conventional industrial platforms. The effective management of such a system relies heavily on the resilience of the network, especially the communication part. Imperative as resilient communication is, there is not enough attention from literature and a standardized framework is still missing. In awareness of these, this paper intends to provide a systematic overview of resilience in IIoT with a communication perspective, aiming to answer the questions of why we need it, what it is, how to enhance it, and where it can be applied. Specifically, we emphasize the urgency of resilience studies via examining existing literature and analyzing malfunction data from a real satellite communication system. Resilience-related concepts and metrics, together with standardization efforts are then summarized and discussed, presenting a basic framework for analyzing the resilience of the system before, during, and after disruptive events. On the basis of the framework, key resilience concerns associated with the design, deployment, and operation of IIoT are briefly described to shed light on the methods for resilience enhancement. Promising resilient applications in different IIoT sectors are also introduced to highlight the opportunities and challenges in practical implementations.
In order to derive safety engagement factors in the workplace and analyze the characteristics of the factors, we collected literature data to be analyzed by a systematic literature review and text mining analysis. We used safety, industrial, occupational, corporate, commitment, engagement, interaction, and participation as key search terms for literature selection and used 143 literature datasets for analysis. We divided the factors of workplace safety engagement into the organizational level and the individual level. In studies after 2005, texts at the individual psychological level appeared in large numbers. Although individual factors have been studied as subfactors at the organizational level, we confirmed that the two types of factors must interact for safety engagement in the workplace. We classified safety engagement factors into cognitive, emotional, behavioral, and relational factors. In particular, relational factors were mainly composed of factors that negatively affected engagement. In the follow-up study, we identified the maturity level among safety engagement factors as divided into four dimensions needed to create a safe workplace environment and to suggest a direction for employees to engage themselves in safety.
Industrial safety. Industrial accident prevention, Medicine (General)
Maleeha MIRHOSSEINI, Mahmoud MOINUDDIN, Forough HIRANI
et al.
Introduction: Ethics is a pervasive subject which covers all aspects of human life. The rapid growth of human society and the complexity of social relations require the emergence of various professions. Survival of these professions depends on the type and quality of services they provide and the trust and credibility that they gain as a result of providing these services. To increase the impact of professional ethics, it is necessary to have patterns consistent with culture and society, and by recognizing them, the dimensions of safe behavior by accountants can be explored.
Research Method: This was an applied study in terms of purpose, and based on the research method, it was both quantitative and qualitative. The statistical population consisted of published domestic papers related to accounting professional ethics. In this study, effective criteria based on previous research were identified and selected in the form of 5 main indicators. Then, a researcher-made questionnaire was designed and implemented to determine the fuzzy cognition mapping pattern.
Findings: Fuzzy cognition mapping among 5 main components showed that there is a cause and effect relationship between all components. However, this cause and effect relationship is positive in some cases and negative in others. Results indicated that an individual component has a negative relationship with an organizational component and a positive relationship with other components. Social component has a positive relationship only with the individual component and a negative relationship with other components. In other words, social component is inversely related to organizational, professionalism and environmental indicators, and the highest intensity of the reverse flow is related to the environmental component.
Conclusion: By understanding professional ethics of accountants and identifying its basic components and determining the relationship between these components in different dimensions and specifying the importance of each of them, a specific framework or format can be designed and implemented for observing or not observing professional ethics by accountants and the desire to behave based on the code of professional behavior .This is to reduce unsafe behaviors, and as a result, reduce the rate of accidents in the country's industries.
Industrial safety. Industrial accident prevention, Public aspects of medicine
Deep learning promises performant anomaly detection on time-variant datasets, but greatly suffers from low availability of suitable training datasets and frequently changing tasks. Deep transfer learning offers mitigation by letting algorithms built upon previous knowledge from different tasks or locations. In this article, a modular deep learning algorithm for anomaly detection on time series datasets is presented that allows for an easy integration of such transfer learning capabilities. It is thoroughly tested on a dataset from a discrete manufacturing process in order to prove its fundamental adequacy towards deep industrial transfer learning - the transfer of knowledge in industrial applications' special environment.
Green Management (GM) is now one of many methods proposed to achieve new, more ecological, and sustainable economic models. The paper is focused on the impact of the developing human population on the environment measured by researched variables. Anthropopressure can have both a positive and a negative dimension. This paper aims to present an econometric model of the Green Industrial Revolution (GIR) impact on the Labour Market. The GIR is similar to the Fourth Industrial Revolution (FIR) and takes place as the next stage in the development of humanity in the perception of both machines and devices and the natural environment. The processes of the GIR in the European Union can be identified based on selected indicators of Sustainable Development (SD), in particular with the use of indicators of the Green Economy (GE) using taxonomic methods and regression analysis. The GM strives to implement the idea of the SD in many areas, to transform the whole economy, and elements of this process are visible Green Labour Market (GLM). The adopted direction of economic development depends on the as-sumptions of strategic management, which can be defined, for example, with green management, which is mainly manifested in the creation of green jobs.
Monalisa Ma'rifat, Atiya Thifal Rofifa, Tri Martiana
Introduction: The plate manufacturing production unit is one of the work units in PT. INKA (Persero), which involves the interaction between humans and machines in its activities, heavy equipment, and materials, all of which can cause possible hazard impacts that can impact the safety and health of workers. The purpose of this study is to conduct risk assessment on occupational safety and health aspects by identifying risks, assessing risks, identifying control efforts and assessing residual risk as a form of efforts to prevent occupational accidents and occupational diseases, using existing resources effectively and efficiently. Method: This research is a type of qualitative research, through interviews and observations, with cross-sectional studies and descriptive analysis. The interviewees for this study were K3LH management managers, steel managers, and machine operators in the plate production unit (PPL). The tools in this study werean interview guide, Job Safety Analysis (JSA) and Hazard Identification Risk Assessment Determining Control (HIRADC) using the AS / NZS 4360: 2004 Risk Management Worksheet Standard Risk Matrix. Results: From the research, it was found that there are 94 hazards for 11 different machines. Regarding the risk levels, there are 9 extreme risk levels, 46 high risk levels, 33 medium risk levels and 6 low risk levels. Conclusion: There are still 61 risks with medium risk level and 6 remaining risks with high risk level that still need control. Control efforts have been implemented by PT. INKA (Persero) in accordance with the hierarchy of control, such as the use of PPE and the provision of work SOPs.
Keywords: hazard identification, risk management, risk assessment, risk control, residual risk