Event Identification for Supply Chain Risk Management Through News Analysis by Using Large Language Models DOI Creative Commons
Maryam Shahsavari, Omar Khadeer Hussain, Morteza Saberi

и другие.

The Review of Socionetwork Strategies, Год журнала: 2024, Номер 18(2), С. 255 - 278

Опубликована: Июль 15, 2024

Abstract Event identification is important in many areas of the business world. In supply chain risk management domain, timely events vital to ensure success operations. One sources real-time information from across world news sources. However, analysis large amounts daily cannot be done manually by humans. On other hand, extracting related depends on query or keyword used search engine and content. Recent advancements artificial intelligence have opened up opportunities leverage intelligent techniques automate this analysis. This paper introduces LUEI framework, a lightweight framework that, with only event’s name as input, can autonomously learn all phrases associated that event. It then employs these for relevant presents results label indicating their relevance. Hence, conducting analysis, able identify occurrence event real The framework’s novel contribution lies its ability those (termed Contributing Events (CEs)) contribute event, offering proactive approach chains. Pinpointing CEs vast data gives managers actionable insights mitigate risks before they escalate.

Язык: Английский

Anomalous Propagation Path Search in Multiplex Networked Industrial Chains DOI
Fulin Chen, Kai Di, Pan Li

и другие.

Опубликована: Янв. 1, 2025

In the context of economic globalization, industrial chains are becoming increasingly complex, with multiple industries and enterprises interwoven to form a multiplex network structure. This complexity may exacerbate propagation anomalies within network, thereby amplifying systemic risks. To effectively address losses caused by anomalous propagation, accurately identifying paths has become critical approach. However, traditional path search methods face two major challenges in chains: (1) existing often assume single-type structure, making it difficult capture implicit relationships between cross-chain nodes; (2) algorithms computationally inefficient cannot handle dynamic features large-scale, multi-layer networks. these issues, this paper proposes an method based on Dynamic Adaptive Genetic Expression Programming (DA-GEP) Biased Restarted Multiple Random Walk (BR-MRW) algorithms. The DA-GEP algorithm adaptively models nodes, helping construct optimize chain BR-MRW simulates interactions aiding discovery global associations. first improves GEP node chains, introducing three adaptive evolution strategies multi-DA-GEP-based construction framework. Additionally, (RW) transition probabilities redefined, new probability matrix computation direction-limiting mechanism information transmission. Experimental results show that outperforms six comparison accuracy, recall, F1 score, average improvements 2.63%, 4.39%, 3.6%, respectively.

Язык: Английский

Процитировано

0
Complexity to Resilience: Machine Learning Models for Enhancing Supply Chains and Resilience in the Middle Eastern Trade Corridor Nations DOI Creative Commons
Wajid Nawaz, Z H Li

Systems, Год журнала: 2025, Номер 13(3), С. 209 - 209

Опубликована: Март 18, 2025

The durable nature of supply chains in the Middle Eastern region is critical, given region’s strategic role global trade corridors, yet geopolitical conflicts, territorial disputes, and governance challenges persistently disrupt key routes like Suez Canal, amplifying vulnerabilities. This study addresses urgent need to predict mitigate chain risks by evaluating machine learning (ML) models for forecasting economic complexity as a proxy resilience across 18 countries. Using multidimensional secondary dataset, we compare gated recurrent unit (GRU), support vector regression (SVR), gradient boosting, other ensemble models, assessing performance via MSE, MAE, RMSE, R2. results demonstrate GRU model’s superior accuracy (R2 = 0.9813; MSE 0.0011), with SHAP, sensitivity, sensitivity analysis confirming its robustness identifying determinants. Analyses reveal infrastructure quality natural resource rents pivotal factors influencing index (ECI), while disruptions embargoes or failures significantly degrade resilience. Our findings underscore importance diversifying investments stabilizing frameworks buffer against shocks. research advances application deep analytics, offering actionable insights policymakers logistics planners fortify regional corridors ripple effects.

Язык: Английский

Процитировано

0
Modelling supply chain risk events by considering their contributing events: a systematic literature review DOI Creative Commons
Maryam Shahsavari, Omar Khadeer Hussain, Pankaj Sharma

и другие.

Enterprise Information Systems, Год журнала: 2025, Номер unknown

Опубликована: Март 24, 2025

Proactive Supply Chain Risk Management (SCRM) helps organisations anticipate and mitigate risks, ensuring business continuity resilience in a violet market. Existing research proposes various techniques to quantify risk occurrence, but none account for the causal relationships between contributing events events. This paper addresses this gap through systematic literature review of SCRM outlines future directions enhance proactive by incorporating dependencies quantification.

Язык: Английский

Процитировано

0
Accelerate demand forecasting by hybridizing CatBoost with the dingo optimization algorithm to support supply chain conceptual framework precisely DOI Creative Commons
Ahmed M. Abed

Frontiers in Sustainability, Год журнала: 2024, Номер 5

Опубликована: Авг. 13, 2024

Supply chains (SCs) serve many sectors that are, in turn, affected by e-commerce which rely on the make-to-order (MTO) system to avoid a risk following make-to-stoke (MTS) policy due poor forecasting demand, will be difficult if products have short shelf life (e.g., refrigeration foodstuffs). The weak negatively impacts SC such as production, inventory tracking, circular economy, market demands, transportation and distribution, procurement. obstacles are data types massive, imbalanced, chaotic. Using machine learning (ML) algorithms solve problem works well because they quickly classify things, makes accurate possible. However, it was found accuracy of ML varies depending sectors. Therefore, presented conceptual framework discusses relations among algorithms, most related sectors, effective scope tackling their data, enables companies guarantee continuity competitiveness reducing shortages return costs. supplied show sales were made at 47 different online stores Egypt KSA during 413 days. article proposes novel mechanism hybridizes CatBoost algorithm with Dingo Optimization (Cat-DO), obtain precise forecasting. Cat-DO has been compared other six check its superiority over autoregressive integrated moving average (ARIMA), long short-term memory (LSTM), deep neural network (DNN), categorical boost (CatBoost), support vector (SVM), LSTM-CatBoost 0.52, 0.73, 1.43, 8.27, 15.94, 13.12%, respectively. Transportation costs reduced 6.67%.

Язык: Английский

Процитировано

2
Event Identification for Supply Chain Risk Management Through News Analysis by Using Large Language Models DOI Creative Commons
Maryam Shahsavari, Omar Khadeer Hussain, Morteza Saberi

и другие.

The Review of Socionetwork Strategies, Год журнала: 2024, Номер 18(2), С. 255 - 278

Опубликована: Июль 15, 2024

Abstract Event identification is important in many areas of the business world. In supply chain risk management domain, timely events vital to ensure success operations. One sources real-time information from across world news sources. However, analysis large amounts daily cannot be done manually by humans. On other hand, extracting related depends on query or keyword used search engine and content. Recent advancements artificial intelligence have opened up opportunities leverage intelligent techniques automate this analysis. This paper introduces LUEI framework, a lightweight framework that, with only event’s name as input, can autonomously learn all phrases associated that event. It then employs these for relevant presents results label indicating their relevance. Hence, conducting analysis, able identify occurrence event real The framework’s novel contribution lies its ability those (termed Contributing Events (CEs)) contribute event, offering proactive approach chains. Pinpointing CEs vast data gives managers actionable insights mitigate risks before they escalate.

Язык: Английский

Процитировано

0