Authors - Shivam Kumar, Dinesh Kumar Saini Abstract - KG-RAG (Knowledge Graph-Retrieval Augmented Generation) is an advanced AI framework that combines structural knowledge graphs with LLMs to make them smarter, more accurate, robust, and less prone to hallucination. However, existing KG-RAG pipelines are often tightly coupled with specific domains. In addition, most of the systems lack proper schema validation and have limited support for temporal knowledge. GenericKG is a modular framework designed to decouple knowledge ingestion, validation, storage and retrieval across domains. The framework includes an agentic ingestion pipeline with schema-driven knowledge graph construction, supported by multi-level validation (L1-L3) to ensure structural, semantic and temporal consistency. Temporal attributes and semantic embeddings are integrated at framework level, enabling time-aware querying and hybrid retrieval without domain-specific reengineering. This paper is evaluated on three benchmarks: the BC5CDR biomedical corpus (87.92% entity F1 with 100% precision), the WebNLG crossdomain dataset (85.6% entity F1 across 15+ relation types on 100 records), and HotpotQA multi-hop question answering (58.0% accuracy on bridge and comparison questions). A raw-LLM baseline without schema guidance scores 0% on all metrics, confirming the importance of the schemadriven pipeline. This framework is implemented in TypeScript and it will be released as open source.
Authors - Abraham Gezehei, Thomas Hanne, Rolf Dornberger Abstract - This study benchmarks twelve recurrent neural network (RNN) architectures for univariate macroeconomic time-series forecasting, covering LSTM and GRU baselines, width/depth scaling, bidirectional encoders, an attention-like pooling variant, convolutional–recurrent hybrids, and strong regularization. Following the Libra benchmarking philosophy and the multi-metric evaluation advocated by Prater et al., we compare all configurations under identical protocols on 100 series from the Libra Economics collection. A bidirectional GRU yields the best RNN accuracy (sMAPE 41.0, MASE 0.0447), improving over a comparable 2-layer GRU baseline (sMAPE 41.9) at higher wall-clock runtime. Most architectural additions and capacity increases do not improve performance over the simple GRU baseline (e.g., deeper/wider models, pooling-based attention, CNN–RNN hybrids, and heavy dropout). The results suggest that short input windows (dynamically sized at 10% of series length, minimum 10 steps) limit the benefits of architectural complexity in this setting. Classical statistical methods (sNaive, ETS, Theta) outperform all neural models by a wide margin while requiring substantially less computation. For these low-frequency macroeconomic series, shallow GRU variants—especially bidirectional encoders—are the strongest RNN option, but classical baselines remain the practical choice.
Authors - Alfito Athar Rayyansyah, Abdurrahman Faris Indriya Himawan, Galuh Sudarawerti Abstract - Governance challenges remain a major concern in large-scale procurement activities, particularly regarding transparency, accountability, and operational effectiveness. This study investigates the role of Big Data Analytics (BDA) and Information Processing Capability (IPC) in enhancing governance outcomes within the e-procurement environment of PT PLN Indonesia Power. Specifically, the study examines how these capabilities contribute to transparency and accountability and how they affect both financial and non-financial procurement performance. A quantitative research design was employed, and data were gathered from employees engaged in procurement-related activities. The proposed model was analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) with SmartPLS 4.0. The findings reveal that all proposed hypotheses are statistically supported. BDA emerged as the primary factor driving transparency and accountability, which subsequently improves procurement performance, particularly non-financial outcomes. The findings reveal that IPC serves as a key enabler in maximizing the value of BDA while increasing the ability of e-procurement systems to support data-driven analysis. These findings offer practical implications for state-owned enterprises by emphasizing the importance of integrating analytical capabilities and information-processing resources to strengthen governance quality and improve procurement effectiveness in digital environments.
Authors - Ichwan Masnadi, Renza Fahlevi, Elda Nurmalinda Abstract - The purpose of this study is to analyze how Perceived usefulness of smart lighting can affect Revisit Intention through the mediation of Green Hotel Image. This study was conducted on hotel guests who stayed at hotels that implemented smart lighting technology in Jakarta. This study uses quantitative methods by sending questionnaires online via Google Forms to 150 hotel guests who have previously stayed at hotels with smart lighting technology implemented. The data was then processed using SEM-PLS (Structural Equation Modeling–Partial Least Square) through SmartPLS 3 software. The results showed that Perceived usefulness of smart lighting had a positive and significant impact on Green Hotel Image. Green Hotel Image also had a positive and significant effect on Revisit Intention. Perceived usefulness of smart lighting had no effect on Revisit Intention. Furthermore, results from the analysis showed that Green Hotel Image fully mediated the effect of Perceived usefulness of smart lighting on Revisit Intention. In conclusion, guests are not inclined to revisit hotels that implement smart technology such as smart lighting. Smart technology indirectly fulfills its role by increasing the hotel’s green (environment-friendly and sustainability focused) image which leads to customer revisit intention. This study contributes to the SOR Theory by showing how Perceived usefulness of smart lighting is the Stimulus factor, Green Hotel Image is the Organism factor, and Revisit Intention is the Response factor. Hotel managers can benefit from this study by properly branding their hotels’ sustainability to leverage their use of smart technology in order to compete with other hotels.
Authors - Luong Vinh Quoc Danh, Truong Minh Nhan, Nguyen Tan Dat, Nguyen Vinh Thanh, Do Chi Tam, Le Tan My, Nguyen Chanh Nghiem Abstract - Accurate avocado ripeness assessment is essential for ensuring product quality and effective postharvest management, yet conventional evaluation methods remain largely destructive, time-consuming, and limited to representative samples. This paper presents a non-destructive ripeness assessment method combining microwave sensing with feedforward neural network (FNN) classification. A custom-designed open-ended coaxial probe connected to a vector network analyzer was employed to measure the complex reflection coefficient S11 of avocado samples over a frequency range of 1.1–3.1 GHz. Variations in the dielectric properties of avocado flesh during ripening produce corresponding and measurable changes in the S11 characteristics, from which magnitude, phase, and frequency features were extracted and used as inputs to the FNN classifier. The proposed system achieved an overall classification accuracy of 87% in discriminating among three ripeness stages – unripe, ripe, and overripe – thereby demonstrating its viability as a rapid, costeffective, and non-destructive alternative to conventional destructive ripeness assessment methods.
Authors - Christian Vasta, Rolf Dornberger, Thomas Hanne Abstract - The Inventory Routing Problem (IRP) is a critical challenge in logistics, combining vehicle routing with inventory management under a unified objective. Recent research in computational intelligence has advanced the use of metaheuristics for tackling such combinatorial problems. Among these, Simulated Annealing (SA) remains underexplored for IRP compared to more commonly applied methods. In this study, we address this gap by implementing a custom SA algorithm to solve a deterministic five-day IRP. The goal is to minimize total transportation costs while satisfying daily customer demand using a single-vehicle fleet with fixed capacity. The algorithm's performance is evaluated with 20 independent runs and compared to a modified Tabu Search benchmark using the same deterministic instance. Our results show that Simulated Annealing performs competitively, producing high-quality solutions, with moderate variation observed across different cooling schedules and repeated runs. Although it shows greater sensitivity to initial parameters and stochastic behavior, its exploratory nature allows it to overcome local optima more effectively than Tabu Search in some cases. The outcomes suggest that SA is a viable alternative for IRP under deterministic conditions, particularly when flexibility in parameter tuning is prioritized.
Authors - Febin Koshy Jacob, Indranil Bose, Sarika D Tavhare, Sandhya Anilkumar Abstract - Modern automotive Electronic Control Unit (ECU) systems demand robust and accurate validation frameworks to address increasing system complexity while minimizing manual test effort and development cost. This paper novels an automated Hardware-in-the-Loop (HIL) testing framework for validation of automotive systems, with a primary focus on automated waveform pattern analysis method. The framework integrates a dSPACE real-time interface with a hardware test bench and algorithm developed using a MATLAB-based simulation model of the Body Control Module (BCM) to generate and analyze input signals. Python-based automation scripts are utilized for test execution control, synchronized data acquisition, and automated result analysis, ensuring repeatable and scalable testing across multiple application domains. The core contribution is a reference-driven waveform comparison methodology, where signals captured from the Device Under Test (DUT) are evaluated against predefined golden reference waveforms. The approach quantifies Root Mean Square Error (RMSE) percentage and timing deviations across individual channels, enabling precise detection of mismatches in waveform sequences. The framework is demonstrated through automotive tail lamp animation pattern validation, where output sequences are compared against reference waveforms for accuracy and robust assessment. Additionally, the solution is extendable to electric vehicle subsystems such as Battery Management Systems (BMS), Traction Motor Control Units (TMCU), and Off-Board Chargers (OFBC), supporting both dynamic and steady-state validation such as torque-speed curve, Battery profile testing, Sensor accuracy etc. The implementation achieves approximately 45.8% automation of test cases and reduces overall validation time by about 41.2%, resulting in improved repeatability, reduced manual intervention, and faster development cycles, ultimately enabling faster time-to-customer and providing a scalable and efficient solution for modern automotive and electric vehicle system validation.
Authors - Apolinar P. Datu, Jeferson C Mojica, Pamela Daphne R. Busog, Kelvin M. Custodio, Desiree Anne D. Mendoza, Kristel Shane C. Paminter, Rose Ann T. Genova, Keno A. Villavicencio Abstract - This study explores how on-the-job training (OJT) helps student interns improve their ability to work with others. It focuses on how real workplace exposure strengthens teamwork, communication, and adaptability. Data were collected from 150 interns from different academic programs using a survey that examined their experiences during training. The findings show that most interns felt a noticeable improvement in their collaborative skills. Many were actively involved in meetings, team activities, and workplace discussions, which gave them valuable opportunities to interact and contribute. These experiences not only helped them communicate more confidently but also made them more comfortable working as part of a team. The results also indicate that supportive work environments—those that encourage communication and teamwork—play an important role in helping interns grow. In addition, OJT helped boost their confidence, sense of responsibility, and readiness for future employment. Overall, the study highlights the importance of OJT as a bridge between academic learning and real-world practice. It reinforces the idea that hands-on experience is essential in preparing students for a workplace that values collaboration and adaptability.
