Abstract International construction disputes frequently involve overlapping jurisdictions, complex legal frameworks, and multifaceted contractual arrangements, posing significant challenges for tribunals in reaching fair and informed decisions. This study examines the role and influence of legal expert witnesses, particularly in construction disputes, through a structured, multistep desktop research methodology. It is guided by two principal objectives: first, to investigate the diverse roles of legal expert witnesses and the significance of their contributions in proving foreign law; and second, to investigate the factors influencing tribunals’ acceptance of legal expert testimony, including the clarity and coherence of such testimony, and credibility concerns relating to the appointing entity, specifically whether tribunals should rely on party-appointed legal experts or whether such concerns render tribunal-appointed experts preferable. The findings indicate that legal experts play a pivotal role in interpreting foreign law, clarifying contractual ambiguities, and providing evidential bases for a tribunal’s determination. However, tribunals’ reliance on expert testimony is conditional on its perceived clarity of testimony as well as the credibility and neutrality of the expert. While party-appointed experts promote adversarial rigor and party autonomy, tribunal-appointed experts enhance independence and institutional confidence. The study concludes that a balanced or hybrid approach provides the most effective mechanism for proving foreign law in international construction disputes. The study further offers recommendations aimed at strengthening the effectiveness, reliability, and legitimacy of legal expert evidence.
Due to the COVID-19 pandemic and the lockdown in early 2020, museums closed their doors and shifted online. Building on recent advancements in digital technologies, they utilized virtual reality (VR) and other technologies to engage visitors and enhance their virtual museum experiences. Nevertheless, museums' VR efforts concentrate on singleton (individual) visits, even though visiting a museum is inherently a social experience. This research illuminates the social dimensions of the virtual museum through a non-holistic approach to explore the impact of social presence on the narrative construction of artifacts in a virtual art museum named "Soul versus Souls," deployed on the Spatial.io metaverse platform. The authors utilized a quasi-experimental approach to control the variables and translate conceptual definitions into operational terms that can be measured and observed. They assessed the museum narrative based on visit style (visitors' movement patterns), visit time (duration of the visit), and visitor narration (self-reflection and non-verbal interactions). The authors employed protocol analysis, observations, and retrospective interviews to document and analyze the findings. The experimental findings suggest that smaller groups facilitate focused conversations, which enhance engagement with the artifacts, while larger groups may introduce distractions that hinder narrative construction.
Objective This study investigated 3D-printed models incorporated with silanized barium sulphate (BaSO4) particles at varying concentrations, focusing on their radiopacity, surface roughness, and microhardness across different printing orientations. Methods BaSO4 was incorporated into a 3D-printed model resin to create four groups: control, 5 wt%, 10 wt%, and 15 wt%. Each group was fabricated at three orientations (0 degrees, 45 degrees, and 90 degrees) (n = 10). Radiopacity was evaluated using cone-beam computed tomography (CBCT) through global histogram analysis. Surface roughness was measured with a contact profilometer, while microhardness was assessed using a Vickers hardness tester. Data were analyzed using two-way ANOVA and Tukey's post hoc test (alpha=0.05). Results Both BaSO4 concentration and printing orientation significantly affected voxel intensity and distribution width (P < 0.001). Specimens printed at 0 degrees consistently exhibited the highest modal intensity values, reaching 85.3% +/- 1.9% of the grayscale spectrum and corresponding to the "Enamel" and "Metal" density ranges. Surface roughness was significantly influenced by all tested factors, with printing orientation showing the greatest effect (P < 0.0001, eta p & sup2; = 0.6). Microhardness increased with increasing BaSO4 content regardless of printing orientation, with filler concentration exerting the strongest influence (P < 0.0001, eta p & sup2; = 0.8). Conclusions An integrated trade-off analysis indicates that a 10 wt% BaSO4 concentration printed at 0 degrees orientation relative to the scan plane provides the optimal clinical configuration. This configuration achieved the "Enamel" and "Metal" segmentation thresholds, produced a high-contrast modal density peak (85.3% +/- 1.9%), maintained adequate microhardness, and minimized image noise, making it the most clinically favorable combination for CBCT-guided applications. Clinical significance Radiopaque 3D-printed models are particularly valuable in CBCT-guided treatment planning and guided implant surgery. Adequate radiographic visibility enables accurate visualization of the printed model relative to the patient's anatomy, facilitating the assessment of implant position, angulation, and depth during virtual planning. Furthermore, radiopaque surgical guides allow verification of guide seating and sleeve alignment on CBCT images, reducing the risk of undetected misfit and subsequent surgical inaccuracies
Abstract Forecasting monetary outcomes in construction delay litigation remains difficult because cost awards are shaped by interacting factual, evidentiary, contractual, and forum-specific considerations rather than by incurred cost alone. This study develops interpretable machine learning models to estimate cost awards from published US construction delay decisions, with separate federal court and state court models to avoid pooling materially different adjudicative environments. The case database was assembled from Google Scholar case law searches using “delay” AND “critical path,” followed by eligibility screening, deduplication, and evidence-traceable coding. The coding framework was first piloted using 33 candidate factors and then reduced to 29 retained factors after excluding sparsely and inconsistently reported variables. Predictors were harmonized using controlled vocabularies, missing categorical information was coded as an explicit Not Specified category rather than imputed by mode, and monetary awards were modeled on a log-transformed scale. Extreme gradient boosting (XGBoost) was selected because it can model nonlinear interactions among heterogeneous categorical predictors while remaining interpretable through global feature importance diagnostics. Using an 80/20 train–test evaluation, the federal model achieved test R 2 of 0.837, root-mean-square error (RMSE) of USD 0.347 million, and mean absolute error (MAE) of USD 0.274 million, while the state model achieved test R 2 of 0.843, RMSE of USD 0.188 million, and MAE of USD 0.132 million. Percentage error metrics are also reported, although mean absolute percentage error (MAPE) and mean percentage error (MPE) are interpreted cautiously where actual awards approach zero. The findings support transparent, forum-specific benchmarking for early case assessment, subject to the limits of published US court decisions and the need for case-specific legal and forensic schedule judgment.
The paper introduces a powerful, data-driven framework for predicting the electrochemical behavior of Polyindole (PIn)-based supercapacitor electrodes through the integration of traditional machine learning (ML) techniques with modern deep learning (DL) architectures. Motivation behind this study comes from the necessity for reliable and accurate models that would enable quick and efficient prediction of the behavior of the conducting polymer electrodes and thus minimize expenses in terms of time, effort, and resources spent on experimental tests needed for characterization of their energy storage capabilities. PIn was prepared through potentiostatic electrochemical polymerization of indole on a graphite sheet. The deposition time was carefully varied to control film growth and active mass loading, and the fabricated electrodes were characterized using cyclic voltammetry (CV) within a potential window from -0.2 to 1.2 V at scan rates ranging from 5 to 75 mV/s. These measurements were then used to assemble a large dataset of over 30,000 experimental points, and provide this dataset to all five ML models Random Forest (RF), K-Nearest Neighbor (KNN), Gradient Boosting (GB), Extreme Gradient Boosting (XGBoost), and Ridge Regression (RR) and four DL networks Artificial Neural Network (ANN), Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), and a hybrid network using RNN-LSTM. The performance of the model was determined by the coefficient of determination (R-2), mean absolute error (MAE), and root mean square error (RMSE). The findings indicate that KNN was best in a high-rate situation (R-2 = 0.9387, MAE = 0.000289, RMSE = 0.000644 when 5 mV/s), whereas ANN was more robust because it sustained a high predictive accuracy (R-2 > 0.9) at all scan rates. Comprehensively, the proposed framework reflects the complementary efficiency of both classical and DL algorithms and enables the application of hybrid deep architectures for reliable predictive modeling of conducting-polymer Supercapacitor (SC) electrodes.
