In recent years, many research has focused on emergency evacuation plans because without them; many lives will be lost. However, mostly evacuation plans aim to lead evacuees to the nearest emergency exit, without taking congestions into account. This nearest exit strategy has led to injuries and casualties on numerous occasions due to stampedes trampling and delays in several instances. In fact, the death toll of the badly implemented evacuation plan exceeds that of the actual reason for the evacuation. Therefore, the development of a Balanced-Evacuation algorithm for Multiple-Exits facilities (BEME) is an important contribution that has the potential to save more lives and to have in emergency situations. The performance of the proposed algorithm was benchmarked against two outstanding–work in the area, including the Depth-First Search (DFS) and Simulated Annealing (SA) algorithms. The results show that BEME algorithm is more efficient in decreasing the maximum crowd at any exit (MCE) and the total evacuation time (TET).
Detecting E-Commerce Fake Arabic Reviews Using Machine Learning Techniques
Research on opinions analysis, one of the areas related to natural language processing, has recently been increasing. Many opinions and thoughts on relevant topics are available online, allowing various parties, such as customers, companies, and even governments, to explore them. E-commerce Arabic reviews shared by online stores are useful for decision-making by customers and e-commerce corporations alike. However, these reviews may be manufactured; fake positive reviews are sponsored by companies to promote their products, and such companies may also give competing products unreasonably negative or indifferent reviews to harm their reputations among consumers. Fake reviews are harmful for both customers and e-commerce corporations and have been acknowledged as critical challenges by the e-commerce industry. To address these issues, this study develops a web-based tool to detect e-commerce fake Arabic reviews by using machine learning algorithms. Naive Bayes and support vector machine algorithms will be used for the classification process as a comparative experiment. An Arabic lexicon will be used to extract some features in order to increase the classifiers’ accuracy. A data collection code (scraper) will be written to collect data. In addition, it will be available for research purposes.
A Trust Management System for Peer-to-peer Network based on Ant Colonies
To be updated
Usability and Security Analysis of the KeepKey Wallet
Hardware Bitcoin wallets store users’ private keys on a secure hardware device to ensure protected and reliable transactions signature. Unlike software or web-based wallets, hardware Bitcoin wallets are immune to malware and can be used securely and interactively. Despite the lauded security of hardware wallets, the security of such devices can be susceptible to the fallibility of human users and the usability of the device. This project conducted an investigation of the usability and security of the hardware wallet KeepKey 4.2.10. A Man-in-the-Middle (MITM) attack was implemented to simulate potential attacks and evaluate its implications for the wallet’s security. In the usability study, ten participants were observed as they used the wallet to complete various Bitcoin address comparison and confirmation tasks. Furthermore, the study examined users’ sentiments towards the transaction process of the KeepKey wallet. The results illustrate that the majority of the participants could detect some but not all MITM attacks. Moreover, there were a considerable number of successful MITM attacks in the experiment. From a usability perspective, our findings show that the Keepkey system has some usability issues that should be tackled to enhance the usability and the security of the wallet. Based on our results and related works in the field, we suggest several usability enhancements to improve the usability and the security of the wallet.
HealthyBlockchain is a project currently being led by Dr. Shada Alsalamah with the MIT Media Lab. The goal of this project is stated as: “Achieving a safe privacy-preserving information sharing environment for individualised care using blockchain-based technology in multiple use cases in the healthcare space.” As part of this group, I contribute to the development of an interface for HealthyBlockchain that can anticipate what users might need from the underlying blockchain systems, along with visualisations, to ensure that the interface elements abide by usability and user experience standards.
A Predictive Gesture-Level Model For Virtual Reality Interactions
The Keystroke-Level Model (KLM) is the simplest model of the Goals, Operators, Methods, and Selection Rules (GOMS) family. KLM computes formative quantitative predictions of task execution time. It is widely used by the Human-Computer Interaction (HCI) community to assess expert user performance with low-fidelity prototypes and to compare designs. Since KLM’s conception, several extensions were reported to accommodate the expanding interaction space (e.g. mobile phones, tablets, and in-vehicular information systems). Nevertheless, current predictive models and extensions are insufficient to model Virtual Reality (VR) applications and its interactive expansion. The purpose of this research is to present a model that extended KLM for VR applications to predict the performance time for various interactions. A systematic bibliographic review will be conducted to determine model operators and unit tasks. The times of these operators will be estimated by a user study to increase its predictive accuracy. The new model will then be empirically evaluated to assess its performance within the context of various application domains.
A Balanced Evacuation Algorithm For Multi-Exit Facilities
Blind Fingerstroke-Level Model
GOMS model is a cognitive modelling technique that is used to predict the user’s performance with a system. KLM model is the simplest version of GOMS models. It predicts the execution time that an expert user needs to complete a system task. Recently, with the development of mobile technologies, GOMS models have been extended. However, available enhancements of GOMS models including KLM model target normal users rather than Visually Impaired (VI) users. In this project, a mobile version of KLM model was analysed and enhanced to support VI interactions with mobile applications. Therefore, three instruments were used to analyse blinds mobile interactions. Each instrument provided different information about their interactions. According to the instruments results, the KLM model was analysed. Thus suitable operators were retained, one operator was removed and new operators were introduced to support blinds mobile interaction. An evaluation experiment was designed to evaluate the proposed enhanced model (Blind FLM). The experiment included two user studies each one served the different goal, and it involved three scenarios of the three top used applications in blind users’ community (WhatsApp, YouTube and Twitter). Twenty participants were involved to find the actual execution times of the three selected scenarios. Then, model predictions were compared to the actual execution times to calculatethe model’s prediction error. Root Mean Square Error (RMSE) was used to calculate the prediction error. The prediction error of Blind FLM model ranged from 1.27 to 3.81 percent with average equalled to 2.36 which is an acceptable prediction error. Finally, Blind FLM was implemented in a simple calculator (Blind FLM Calculator) to facilitate the model calculations process for the designers.
Fish-Inspired Task Allocation Approach For Multi- Unmanned Aerial Vehicle Search And Rescue Missions
Carrying out a search and rescue mission (SAR) with multiple unmanned aerial vehicles (UAVs) introduces a grand challenge to the researchers. Answering the question of how to efficiently allocate multiple UAVs to search and rescue tasks within a strict time frame and in a hostile environment is not an easy job. The task allocation problem is classified as NP-hard which increases in complexity as the numbers of UAVs and tasks increase. Therefore, such a problem is usually approached by heuristic techniques, such as market-based and biologically-inspired approaches. This thesis proposes a fish school-inspired algorithm for the multi-UAV task allocation problem in SAR missions. Three long-established task allocation mechanisms, which are the max-sum, auction-based and opportunistic coordination scheme were used to benchmark the proposed algorithm performance. Experimental results demonstrated that the new algorithm yielded a high net throughput and a short mean rescue time while maintaining a linear running time when compared to the benchmarks.
Context-Aware Gossip-Based Protocol For Internet Of Things Applications
This project proposes a gossip-based protocol that utilises a multi-factor weighting function (MFWF) that takes several parameters into account: residual energy, Chebyshev distances to neighbouring nodes and the sink node, node density, and message priority. The effects of these parameters were examined to guide the customisation of the weight function to effectively disseminate data into three types of IoT applications: critical, bandwidth-intensive, and energy-efficient applications. The performances of the three resulting MFWFs were assessed in comparison with the performances of the traditional gossiping protocol and the Fair Efficient Location-based Gossiping (FELGossiping) protocol in terms of end-to-end delay, network lifetime, rebroadcast nodes, and saved rebroadcasts. The experimental results demonstrated the proposed protocol’s ability to achieve a much shorter delay for critical IoT applications. For bandwidth-intensive IoT application, the proposed protocol was able to achieve a smaller percentage of rebroadcast nodes and an increased percentage of saved rebroadcasts, i.e. better bandwidth utilisation. The adapted MFWF for energy-efficient IoT application was able to improve the network lifetime compared to that of gossiping and FELGossiping. These results demonstrate the high level of flexibility of the proposed protocol with respect to network context and message priority.
Bee-Inspired Task Allocation Algorithm For Multi-UAV Search And Rescue Missions
Task allocation plays a pivotal role in the optimization of multi-unmanned aerial vehicle (multi-UAV) search and rescue (SAR) missions in which the search time is critical and communications infrastructure is unavailable. These two issues are addressed by the proposed BMUTA algorithm, a bee-inspired algorithm for autonomous task allocation in multi-UAV SAR missions. In the proposed approach, UAVs dynamically change their roles to adapt to changing SAR mission parameters and situations by mimicking the behaviour of honeybees foraging for nectar. The proposed algorithm is characterized by a complete absence of centralized control and direct communications between UAVs. Four task allocation heuristics (namely, auction-based, max-sum, ant colony optimization (ACO) and opportunistic task allocation (OTA)) were thoroughly tested in simulated SAR mission scenarios to comparatively assess their performance relative to that of BMUTA. The experimental results demonstrate BMUTA’s ability to achieve a superior number of rescued victims in much shorter rescue times and runtime intervals. The proposed approach demonstrates a high level of flexibility based on its situational awareness, high autonomy, and economic communication scheme.
A Multi-UAV Task Allocation Algorithm For Red Palm Weevil Combat Based On Bacteria Behaviour
The red palm weevil (RPW) is spreading over palm trees in many countries, causing tremendous economic losses estimated at multi-million dollars annually. There is a dire need to fight this fatal insect scourge by halting its further spread. A team of multiple unmanned aerial vehicles (UAVs) with suitable equipment has the potential to help by cooperating in RPW detect- and-treat missions. However, a significant challenge arises in this context regarding how to distribute UAVs efficiently among search and detect tasks within the mission’s constraints. This paper proposes UTARB, a multi-UAV task allocation algorithm for RPW combat inspired by bacteria foraging behaviour. The proposed algorithm has been tested thoroughly in simulated detect-and-treat missions to examine its efficiency against two long-standing task allocation algorithms: auction-based and opportunistic task allocation heuristics. Experimental results demonstrated the superiority of the proposed algorithm against the benchmark algorithms, in terms of increased detection rates and reduced mission completion and running times.
An AR Application for the Reading Development of Arabic Students with Hearing Impairments
Hearing is one of the five essential senses for humans. Disruptions of this sense have a negative effect in all aspects of a person’s life including speech, communication, and learning. Literacy is affected by hearing loss and deafness. Deaf children learn how to read by associating a Sign Language (SL) sign with an image and a printed word. Arabic deaf children face reading difficulties due to their weak linguistic crop in ArSL. Despite several researches that aim to address the literacy problem of deaf people, the majority of the work is direct to SL for other languages, such as American and Malaysian SL. This project aims at facilitating the reading process for Arab deaf children, especially students in elementary school. In this research, we will utilize Augmented Reality (AR) technologies to augment current practice for reading. The developed application “كلمة وإشارة” aims to reinforce the reading experience of Arab deaf children by associating printed text with several media, such as videos, sounds, SL, and fingerspelling. Based on preliminary investigation and gathered requirements, we implement an AR mobile application. In addition, We perform a statistical analysis of data gathered from deaf students. It is proved that there is a positive improvement in vocabulary learning of students whenever they use an AR application compared with their counterparts who use the traditional learning method.
A Comprehensive Survey Of Extensions To The Keystroke-Level Model
The Keystroke-Level Model (KLM) is the simplest model of the Goals, Operators, Methods, and Selection Rules (GOMS) family. KLM computes formative quantitative predictions of task execution time. This paper provides a systematic literature review of KLM extensions across various applications and setups. The objective of this review is to address research questions concerning the development and validation of extensions. A total of 54 KLM extensions have been exhaustively reviewed. The result show that the original keystroke and mental act operators were continuously preserved or adapted, and the drawing operator was used the least. Excluding the original operators, almost 45 operators were collated from the primary studies. Only half of the studies validated their model’s efficiency through experiments. The results also identify several research gaps, such as the shortage of KLM extensions for post-GUI/WIMP interfaces. Based on the results obtained in this work, this review finally provides guidelines for researchers and practitioners.
Effect Of Exit Placement On Evacuation Plans
Disasters caused by human crowding presents a dangerous problem, which requires the attention of researchers to determine the smooth flow of pedestrians to ensure safe exits form dangerous situations. In this project, the placement of multiple exits (4 exits) in four different positions is investigated to determine its impact on maximum crowds at exits and evacuation time. These positions were: all on one side, on adjacent sides, on opposite sides, and on all sides. An evacuation system was implemented in an environment of congested pedestrians to comparatively assess the performance of Depth-First Search (DFS) and Simulated Annealing (SA) as the positions of the exits were changed. The simulation results demonstrate that the work placement of exits was when they were on adjacent walls. The SA techniques also proved superior to that of the DFS technique as it better-balanced evacuations. This type of research can positively impact the placement of exits during architectural planning and design to reduce pedestrian crowding while exiting a dangerous event.
A Tangible User Interface For Interactive Data Visualisation
Information visualisation (infovis) tools are integral for the analysis of large abstract data, where interactive processes are adopted to explore data, investigate hypotheses and detect patterns. New technologies exist beyond post-windows, icons, menus and pointing (WIMP), such as tangible user interfaces (TUIs). TUIs expand on the affordance of physical objects and surfaces to better exploit motor and perceptual abilities and allow for the direct manipulation of data. TUIs have rarely been studied in the field of infovis. The overall aim of this thesis is to design, develop and evaluate a TUI for infovis, using expression quantitative trait loci (eQTL) as a case study. The research began with eliciting eQTL analysis requirements that identified high- level tasks and themes for quantitative genetic and eQTL that were explored in a graphical prototype. The main contributions of this thesis are as follows. First, a rich set of interface design options for touch and an interactive surface with exclusively tangible objects were explored for the infovis case study. This work includes characterising touch and tangible interactions to understand how best to use them at various levels of metaphoric representation and embodiment. These design were then compared to identify a set of options for a TUI that exploits the advantages of touch and tangible interaction. Existing research shows computer vision commonly utilised as the TUI technology of choice. This thesis contributes a rigorous technical evaluation of another promising technology, micro-controllers and sensors, as well as computer vision. However, the findings showed that some sensors used with micro-controllers are lacking in capability, so computer vision was adopted for the development of the TUI. The majority of TUIs for infovis are presented as technical developments or design case studies, but lack formal evaluation. The last contribution of this thesis is a quantitative and qualitative comparison of the TUI and touch UI for the infovis case study. Participants adopted more effective strategies to explore patterns and performed fewer unnecessary analyses with the TUI, which led to significantly faster performance. Contrary to common belief bimanual interactions were infrequently used for both interfaces, while epistemic actions were strongly promoted for the TUI and contributed to participants’ efficient exploration strategies.
An Evaluation Of Extended Duration Multi-Touch Interaction
The goal of this project was to evaluate the extended use of multi-touch interaction techniques, more specifically the ergonomic convenience of existing bimanual and unimanual interaction techniques and personal preference over an extended period of time for both horizontal and vertical tabletops. Objective localised muscle fatigue, muscle activity, and subjective perceived exertion measures were administrated. In the experimental design, electromyograms were recorded during tabletop interaction technique and voluntary isometric contractions were recorded pre- and post-tabletop activity for the biceps brachii, middle deltoid, and extensor digitorum for both sides of the body. Changes in the median power frequency (MPF) and root mean square (RMS) were explored to examine muscular fatigue and activity respectively. MPF was found sensitive to fatigue for some muscles on both the horizontal and vertical condition where a decline in MPF was noted, albeit statistically insignificant. Perceived exertion ratings have shown an increase by the end of the task where the difference between the means was found to be statistically significant for the vertical condition but not the horizontal one. The electromyograms recordings, along with video recordings, have shown the sustainability of the interaction techniques adopted at the beginning of the task to the end of that task, which included both unimanual and bimanual techniques.