Nonconformities are the major sources of waste in manufacturing process. Nonconformities cannot be fully eliminated but their occurrence rate can be predicted. This paper proposes a hybrid approach based on ontological modelling and machine learning for predicting the non-conformance rates of a manufacturing process and minimising its associated costs. Based on the proposed approach, the work orders, that are represented semantically using a formal ontology, are first clustered according to their semantic similarities and then, for each cluster, the appropriate models that predict the probability distribution of non-conformance rates are developed. When a new work order is created, the most similar work order is retrieved from historical records, and the probability distribution of its non-conformance rate is estimated by applying the predictive model of the cluster to which the work order belongs. The probability distribution is used to calculate the expected underproduction and overproduction cost and to determine the amount of production that minimises the expected costs. The proposed method was validated using a dataset obtained from a manufacturer of packaging for cosmetics. Compared to the expert’s opinions and other machine learning algorithms, the proposed method demonstrated better performance with respect to cost reduction.

Slope displacement monitoring is essential for assessing slope stability and preventing catastrophic failures, particularly in geotechnically sensitive areas. However, continuous data collection is often disrupted by environmental factors, sensor malfunctions, and communication issues, leading to missing data that can compromise analysis accuracy and reliability. This study addresses these challenges by evaluating advanced machine learning models—SAITS, ImputeFormer, and BRITS (Bidirectional Recurrent Imputation for Time Series)—for missing data imputation in slope displacement datasets. Sensors installed at two field locations, Yangyang and Omi, provided high-resolution displacement data, with approximately 34,000 data points per sensor. We simulated missing data scenarios at rates of 1%, 3%, 5%, and 10%, reflecting both random and block missing patterns to mimic realistic conditions. The imputation performance of each model was evaluated using Mean Absolute Error, Mean Squared Error, and Root Mean Square Error to assess accuracy and robustness across varying levels of data loss. Results demonstrate that each model has distinct advantages under specific missingness patterns, with the ImputeFormer model showing strong performance in capturing long-term dependencies. These findings underscore the potential of machine learning-based imputation methods to maintain data integrity in slope displacement monitoring, supporting reliable slope stability assessments even in the presence of significant data gaps. This research offers insights into the optimal selection and application of imputation models for enhancing the quality and continuity of geotechnical monitoring data.

Slope failures pose significant threats to human safety and vital infrastructure. The urgent need for the accurate prediction of these geotechnical events is driven by two main goals: advancing our understanding of the underlying geophysical mechanisms and establishing efficient evacuation protocols. Although traditional physics-based models offer in-depth insights, their reliance on numerous assumptions and parameters limits their practical usability. In our study, we constructed an experimental artificial slope and monitored it until failure, generating an in-depth displacement dataset. Leveraging this dataset, we developed and compared prediction models rooted in both statistical and machine learning paradigms. Furthermore, to bridge the gap between generic evaluation metrics and the specific needs of slope failure prediction, we introduced a bespoke performance. Our results indicate that while the statistical approach did not effectively provide early warnings, the machine learning models, when assessed with our bespoke performance metric, showed significant promise as reliable early warning systems. These findings hold potential to fortify disaster prevention measures and prioritize human safety.

In the Architecture, Engineering, and Construction (AEC) industry, virtual environments are being utilized to enhance communication among stakeholders and improve visual comprehension. However, stakeholders possess diverse personal characteristics which can affect their spatial recognition ability in virtual spaces. Despite the potential impact of these individual traits, related research still needs to be more comprehensive. Therefore, this study analyzed how each individual’s characteristics influence spatial recognition in a Building Information Model (BIM)-based virtual environment. A quantitative methodology via a survey was employed to investigate the influence of personal factors such as age, gender, education level, and gaming experience on spatial recognition. In a 3D virtual corridor using BIM software, 76 participants were asked to navigate the corridor using a controller and count 23 sprinklers. Of the 76 participants, 30 responses were selected for the statistical analysis. The results demonstrate that age, gender, and education level did not significantly affect spatial recognition in the virtual environment. Conversely, participants with gaming experience tended to perceive spaces in the virtual environment more accurately and realistically, showing a statistically significant difference. This outcome suggests that gaming experience is crucial in enhancing spatial recognition ability in virtual environments. The findings from this study offer critical insights into the impact of individual characteristics on spatial recognition, providing valuable information for the future practical use of BIM-based virtual environments, and can subsequently assist in discovering efficient communication methods among stakeholders.

The engineering properties of asphalt binders depend on the types and amounts of additives. However, measuring engineering properties is time-consuming, requires technical expertise, specialized equipment, and effort. This study develops a deep regression model for predicting the engineering property of asphalt binders based on analysis of atomic force microscopy (AFM) image analysis to test the feasibility of replacing traditional measuring estimate techniques. The base asphalt binder PG 64-22 and styrene–isoprene–styrene (SIS) modifier were blended with four different polymer additive contents (0%, 5%, 10%, and 15%) and then tested with a dynamic shear rheometer (DSR) to evaluate the rheological data, which indicate the rutting properties of the asphalt binders. Different deep regression models are trained for predicting engineering property using AFM images of SIS binders. The mean absolute percentage error is decisive for the selection of the best deep regression architecture. This study’s results indicate the deep regression architecture is found to be effective in predicting the G*/sin δ value after the training and validation process. The deep regression model can be an alternative way to measure the asphalt binder’s engineering property quickly. This study would encourage applying a deep regression model for predicting the engineering properties of the asphalt binder.