IFIP Open Access Digital Library
Not a member yet
22614 research outputs found
Sort by
Postcolonial 4IR Environmental Scanning for IS Education: A Transformative Mixed-Methods Approach for Decolonization and Inclusivity
No full textPart 5: ICT Curriculum and EducationInternational audienceThe Fourth Industrial Revolution (4IR) heralds a significant turning point for Information Systems (IS) education in the global South, particularly in postcolonial nations such as South Africa. This study aims to introduce a preliminary transformative mixed-methods approach to environmental scanning for assessing and responding to the rapidly evolving postcolonial IS educational landscape in the 4IR era. Drawing from recent theoretical and methodological advances in IS and postcolonial research, this approach uniquely combines quantitative data analysis for broader trend identification with qualitative inquiries to capture in-depth perspectives from diverse educational stakeholders in postcolonial contexts. This transformative sociotechnical tool can better align IS curricula with rapid technological advancements and varied industry needs while critically addressing the social, political, cultural, ethnic, and economic development complexities inherent in postcolonial contexts. The paper proposes a complementary approach for employing an adapted environmental scanning approach that extends upon conventional mixed-methods approaches by explicitly addressing social injustice, inequality, and power issues in postcolonial regions. The paper also discusses the sociotechnical challenges of implementing this approach in diverse and resource-constrained educational environments, such as limited funding, access to technology, and a deficit of postcolonially-sensitive educators. This paper aims to contribute to advancing IS education and guiding administrators, educators, and policymakers in developing technologically current and socially inclusive curricula. This paper lays the groundwork for future empirical research and transformative IS education discourse and actions for postcolonial regions in the 4IR era
Artificial Intelligence Applications and Innovations: 20th IFIP WG 12.5 International Conference, AIAI 2024, Corfu, Greece, June 27–30, 2024, Proceedings, Part IV
No full textInternational audienceBook Front Matter of AICT 71
From Tweets to Reddit: Leveraging Semi-supervised Domain Adaptation for Improving Data Filtering
No full textPart 4: LearningInternational audienceReddit has emerged as a leading platform for microblogging data collection, providing valuable insights into patterns and knowledge discovery. However, the process of gathering and preparing this data presents significant challenges, particularly when it comes to ensuring its accuracy. Existing methodologies often yield an abundance of irrelevant posts, and datasets for relevance prediction on Reddit are rare. To overcome these obstacles, we propose a new semi-supervised framework that filters Reddit posts based on their topic relevance. Our approach combines annotated data from Twitter with weak labels generated from Wikipedia pages associated with relevant subreddits to automatically label Reddit posts. To enhance the model’s generalization performance, we utilize a domain adversarial adaptation network to bridge the distribution gap between Twitter and Reddit data. Our novel framework achieves an accuracy of 73% and an F1 score of 0.77, which is a significant improvement of 20% compared to baseline models. Additionally, we address important research questions regarding the effectiveness of automatic labeling, the use of weakly labeled data, the contextual requirements for training domain adaptation models, and the optimal weak labeling method
Integrating LLMs in the Engineering of a SAR Ontology
No full textPart 5: Multi Agent/Ontologies/RoboticsInternational audienceIn Search and Rescue (SAR) missions, the integration of multiple sources of information may enhance operational efficiency and increase responsiveness significantly, improving situation awareness and aiding decision-making to save lives and mitigate incident impact. Ontologies are crucial for integrating and reasoning with data from diverse sources. Engineering a domain ontology for SAR can be better supported from an agile, collaborative, and iterative ontology engineering methodology (OEM), incorporating the interests of several stakeholders. Large Language Models (LLMs) can play a significant role in completing OEM processes. The goal of this work is to identify how ontology engineering (OE) tasks can be completed with the collaboration of LLMs and humans. The objectives of this paper are, a) to present preliminary exploration of LLMs to generate domain ontologies for the modeling of SAR missions in wildfire incidents b) to propose and evaluate an LLM-enhanced OE approach. In overall, the main contribution of the work presented in this paper is the analysis of LLMs capabilities to ontology engineering, and the evaluation of the synergy between humans and machines to efficiently represent knowledge, with specific focus in the SAR domain
Path Planning Optimisation for Multiple Drones: Repositioning the Starting Point
No full textPart 3: Explainable AI - OptimizationInternational audienceDrones’ power management, and its impact on their applications, faces significant challenges. Path planning, and particularly travelled distance optimisation, has been shown to be a significant factor to enhance their efficiency and effectiveness. Given a scenario where a drone must visit static stations, analysis of the significance of a path’s constituent parts shows that the segments from the launch pad and the first/last station to be visited emerged as the primary candidates for further optimisation. Accordingly, and provided the launch pad’s ability to relocate, a number of alternative approaches are herein proposed as candidates for the adjustment of the launch pad in order to minimise the total distance of multiple drones’ paths. Extensive experimentation taking into consideration multiple scenarios pertaining to varying number of stations and available drones indicates that the reposition of the launch pad as center for all drones’ routes, obtained significant improvements/minimisation in the total distance of the path ranging from 4% to 22%
Dynamic Stacking Optimization in Unpredictable Environments: A Focus on Crane Scheduling
No full textPart 3: Explainable AI - OptimizationInternational audienceDynamic Stacking Optimization holds significant practical relevance for multimillion-dollar industries involved in production and delivery processes. It entails the utilization of cranes to relocate products, with the relocation needing to be scheduled while adhering to various time constraints. This paper addresses the challenge of developing solution approaches for such dynamic stacking problems in uncertain environments, particularly the environment represented by the “CraneScheduling" simulation of the DynStack Competition 2023, which is part of the Genetic and Evolutionary Computation Conference. Leveraging systematic and task-oriented solutions enabled by industrial digitalization, our work focuses on optimal crane scheduling to prevent delivery errors, maximize block handling efficiency, and minimize truck queues and shipment delays. The dynamic stacking problem involves two cranes operating on the same girder, multiple arrival and handover stacks, and various stacks in the buffer area. Our approach employs a role-based solver, efficiently planning crane assignments to manage container movements within the terminal. The performance of the solvers is compared with other solvers submitted in the competition, building on existing solution approaches. The role-based solver leads to reduced unnecessary movements by efficiently planning crane assignments. However, it shows potential for significant improvement, offering benefits for both the industry and the environment
A Machine Learning Approach for Points of Interest Extraction and Event Classification
No full textPart 2: Recommendation/ClassificationInternational audienceThis paper presents a novel approach that utilizes machine learning techniques, specifically clustering algorithms and artificial neural networks, to improve the prediction and understanding of human routines in urban mobility contexts. Our method focuses on the identification and categorization of Points of Interest (POIs) from travel data, facilitating the accurate prediction of user routines for intelligent transportation systems. By integrating a clustering phase that groups individual stop points into POIs, followed by a correction mechanism through user interaction, we address the limitations of existing methods in adapting to dynamic mobility patterns and the contextual ambiguity of GPS coordinates. Subsequently, a classification phase employs a feed-forward neural network to assign incoming travel events to the identified POIs. This dual-phase approach not only improves the precision of routine predictions but also enhances the adaptability of the system to changes in mobility behavior over time. The incorporation of a cognitive module, based on Dynamic Neural Fields (DNF), further allows for personalized predictions regarding the timing, duration, and nature of trips. Validated with datasets from the Portuguese city of Braga, our results demonstrate the effectiveness of this methodology in providing actionable insights for the development of cognitive solutions for the project BE.Neutral’s innovative vehicle “BEN”. By emphasizing user involvement and algorithmic transparency, our work contributes to the advancement of smart transportation technologies in urban environments
All Your LLMs Belong to Us: Experiments with a New Extortion Phishing Dataset
No full textPart 5: ML Attack, VulnerabilityInternational audienceIn the last decade, there has been a dramatic rise in phishing emails including business email compromise, and extortion attacks. Ransomware and blackmail are examples of extortion threats, where attackers force victims to follow orders, send money, or share sensitive information. To our knowledge, no phishing dataset that includes text-based extortion attacks has been made available to the security research community. To address this problem, we present “TExtPhish” consisting of a sentence-level subset and a full-body email-level subset that can be used for multiple classification and regression tasks. We also provide another challenging subset with homograph text perturbations to address a specific Unicode NLP attack targeting primarily LLMs, causing them to hallucinate and significantly degrade their performance. We show this by conducting multiple experiments including extortion classification, sentiment analysis, and language identification. Our findings indicate that DistilBERT is most susceptible to homograph attacks at sentence level, resulting in a 94.9% decrease in F1-score while DeBERTa’s performance decreased 94.1% at email level
Human Digital Twins: Efficient Privacy-Preserving Access Control Through Views Pre-materialisation
No full textPart 1: Access ControlInternational audienceDigital Twins (DTs) are virtual copies of physical entities, processes, or systems used for various tasks, such as controlling, monitoring, and analysing the status of the real entity. The DT sector is expected to surpass six billion U.S. dollars by 2025, with the Human Digital Twin (HDT) being a prime example. HDTs are being used in various applications, such as personalised medicine, healthcare, and education. However, the materialisation of HDTs can be costly and lead to delays in HDT-based services. To overcome this, we propose a strategy, HDT-ViewMat, to identify the portions of an HDT that should be pre-materialised, considering the trade-off between potential delays and resource waste. The proposed strategy analyses the process/workflow that requires HDT data to estimate the probability of its tasks being executed. Furthermore, due to the sensitivity of the data maintained by the HDTs, access to them must be limited to guarantee the users’ privacy. This strategy also considers the compliance of privacy policies with users’ preferences. HDT-ViewMat assesses the user’s chance of executing a task in the workflow based on the probability of the task’s invocation and the probability of the user accepting the policies of the corresponding service provider
Online Gambling in the Rural Global South: Probably the Next Major Silent Killer
No full textInternational audienceThis paper combines the author’s evaluative datasets with nascent literature on rural gambling, to come to a narrative about a worsening epidemic of online gambling disorders in the rural Global South. The paper posits that the rapid spread of mobile money enabled gambling companies to penetrate ever deeper into regions they could not previously access. This has generated a significant and growing flow of money from rural regions to distant companies, and a wave of gambling disorders among rural men. There is evidence that online gambling is increasingly affecting rural women as well. The offer is enormous and it is available 24/7, from the privacy of one’s home. Regulation is rudimentary and restrictions are ineffective. In most rural regions harm-reduction support is not available. Online gambling is likely to be the next major silent killer in the rural Global South