Search

Metaheuristics are optimization algorithms that efficiently solve a variety of complex combinatorial problems. In psychological research, metaheuristics have been applied in short-scale construction and model specification search. In the present study, we propose a bee swarm optimization (BSO) algorithm to explore the structure underlying a psychological measurement instrument. The algorithm assigns items to an unknown number of nested factors in a confirmatory bifactor model, while simultaneously selecting items for the final scale. To achieve this, the algorithm follows the biological template of bees' foraging behavior: Scout bees explore new food sources, whereas onlooker bees search in the vicinity of previously explored, promising food sources. Analogously, scout bees in BSO introduce major changes to a model specification (e.g., adding or removing a specific factor), whereas onlooker bees only make minor changes (e.g., adding an item to a factor or swapping items between specific factors). Through this division of labor in an artificial bee colony, the algorithm aims to strike a balance between two opposing strategies diversification (or exploration) versus intensification (or exploitation). We demonstrate the usefulness of the algorithm to find the underlying structure in two empirical data sets (Holzinger-Swineford and short dark triad questionnaire, SDQ3). Furthermore, we illustrate the influence of relevant hyperparameters such as the number of bees in the hive, the percentage of scouts to onlookers, and the number of top solutions to be followed. Finally, useful applications of the new algorithm are discussed, as well as limitations and possible future research opportunities.

Purpose: Course evaluations are formative and are used to evaluate learnings of the students for a course. Anomalies in the evaluation process can lead to a faulty educational outcome. Learning analytics and educational data mining provide a set of techniques that can be conveniently applied to extensive data collected as part of the evaluation process to ensure remedial actions. This study aims to conduct an experimental research to detect anomalies in the evaluation methods. Design/methodology/approach: Experimental research is conducted with scientific approach and design. The researchers categorized anomaly into three categories, namely, an anomaly in criteria assessment, subject anomaly and anomaly in subject marks allocation. The different anomaly detection algorithms are used to educate data through the software R, and the results are summarized in the tables. Findings: The data points occurring in all algorithms are finally detected as an anomaly. The anomaly identifies the data points that deviate from the data set's normal behavior. The subject which is consistently identified as anomalous by the different techniques is marked as an anomaly in evaluation. After identification, one can drill down to more details into the title of anomalies in the evaluation criteria. Originality/value: This paper proposes an analytical model for the course evaluation process and demonstrates the use of actionable analytics to detect anomalies in the evaluation process.

This paper describes Alkaline, a size-reduced version of Kyber, which has recently been announced as a prototype NIST standard for post-quantum public-key cryptography. While not as simple as RSA, I believe that Alkaline can be used in an undergraduate classroom to effectively teach the techniques and principles behind Kyber and post-quantum cryptography in general. Classroom experiences with individual concepts used in Alkaline support this belief. In addition to cryptography, linear algebra and abstract algebra classes would be good candidates for the use of Alkaline. A few exercises suitable for use in these classes are included.

In this paper, several research topics are presented that provide productive avenues of exploration for students in an undergraduate cryptography course. The topics are intended for a variety of skill levels and interests. In some cases, a student might be inspired to try to solve the proposed problem or write computer code. Other topics might motivate them to learn new mathematics or algorithms, or research a historical event or person. In all cases, a short description of the topic is given along with several references that can be used as a starting point for student research. The author also describes some challenges and recommendations when teaching a course where a project is a main component.

The Miller-Rabin test is a useful probabilistic method for finding large primes. In this paper, we explain the method in detail and give three variations on this test. These variations were originally developed as student projects to supplement a course in error correcting codes and cryptography.

The artificial intelligence methods might be applied to see through the education problems, and make effective prediction and decision. The transformation from data to decision are inseparable from the learning analytics. In order to solve the dynamic multi-objective decision problems, a decision learning algorithm is designed to analyze the learning behavior and learning achievement based on the particle swarm optimization mechanism, which ensures the global optimization of data analysis. Compared with the performance indicators of several approximate algorithms, this algorithm has the advantage of association analysis, which improves the accuracy and robustness; Furthermore, three tested problems are constructed by GeoDa tool, and through data training and analysis, the dynamic multi-objective decision rules are obtained. The whole research can provide theoretical and technical support for other related or similar issues.

The last decade has seen a rise in big data approaches, including in the humanities, whereby large quantities of data are collected and analysed. In this paper, we discuss long-form audio recordings that result from individuals wearing a recording device for many hours. Linguists, psychologists and anthropologists can use them, for example, to study infants' or adults' linguistic behaviour. In the past, recorded individuals and communities have resided in high-income countries (HICs) almost exclusively. Recognising the need for better representation of all cultures and linguistic experiences, researchers have more recently started to collect long-form audio recordings in low- and middle-income countries (LMICs). We aim to help researchers to collect, analyse and use these recordings ethically. To do so, we identify four main ethical challenges linked to research that relies on long-form recordings in LMICs. We provide recommendations to overcome these challenges. These considerations should be useful to researchers employing other big data techniques collected via wearables.

A key feature of embodied education is the participation of the learners' body and mind with the environment. Yet, little work has been done to review the state of embodied education with Artificial Intelligence (AI). The goal of this systematic review is to examine the state of human and AI's triad engagement in education, that is the mind, body, and environment. Through a review of N = 38 articles retrieved from SCOPUS and Web of Science (WoS), we code and analyze which of mind, body, and environment is present in each engagement reported per reviewed study. Further, we examine which one of the technologies (which may include AI), human or human + technology is present in each engagement. We summarize the demographic and embodied trends in the reviewed studies. Findings of our review show among the body, mind, and environment triad, the mind is most significantly present in the studies. The reviewed studies are most often concerned with the technicality of embodied AI in education, concentrating on the algorithms and accuracy of facial expressions, speech, etc. Far less attention has been paid to other important learner needs. The contribution of this work is in presenting a blueprint for current research on embodied AI in education, identifying implications of research, and offering a classification that includes the environment-body-mind triad and three possible entities per triad, namely the human, technology/AI, or human + technology. Future work needs to examine the combinations in which the engagement triad and entities may be present and their impact on humans' well-being and research in the field overall.

Various studies have been conducted on applying intelligent recognition technology, especially speech recognition technology to improve English learning ability, mostly listening and speaking. However, few studies have touched on how image-to-text recognition technology can be used for writing. The present research was conducted to fill this gap by exploring the optimization of a deep-learning-based image captioning algorithm to facilitate English writing, so as to enable learners to break the time and space limitations and learn English writing (including sentence patterns, spelling, vocabulary, and grammar) anytime and anywhere by taking pictures. Therefore, this paper focused on image captioning based on CNN(Convolutional Neural Networks) and LSTM (Long Short-Term Memory), using DenseNet201 or Vision Transformer trained on the ImageNet-1K image classification dataset as the image encoder and LSTM as the decoder. First, pre-training was performed on the Flickr8k dataset. After selecting the best-trained model as the pre-trained weight model for the COCO dataset, fine-tuning optimization was performed on the COCO dataset, and the attention mechanism was used to design the ablation experiment. The BLEU-4, CIDEr, METEOR, and ROUGE evaluation indexes of the optimized model on the test set were 0.3437,1.121,0.2750, and 0.5117, respectively. The study results showed that the convergence of the model was accelerated and had better performance. The model was used to automatically caption 12 images that had never been used during the training process. The descriptions generated by the optimized image captioning algorithm have lexical and syntactic accuracy, and matched what the images expressed, showing that this improved algorithm could be used as a learning tool to help English learners improve lexical and syntactic acquisition to promote writing through the generated descriptions of the pictures taken anytime and anywhere in real-life situations.

Knowledge tracing (KT) for evaluating students' knowledge is an essential task in personalized education. More and more researchers have devoted themselves to solving KT tasks, e.g., deep knowledge tracing (DKT), which can capture more sophisticated representations of student knowledge. Nonetheless, these techniques ignore the reconstruction of the observed input information. Therefore, this leads to poor predictions of students' knowledge, even if the student performed well in the past knowledge state. In this article, we first employ causal inference for explanatory analysis of KT, then propose a learning algorithm for stable KT based on the analysis outcomes. The proposed approach aims to achieve stable KT by constructing global balanced weights that facilitate estimating feature influence and assessing causal relationships between individual variables and outcome variables. We have proved the approach has effective in accuracy and interpretability through extensive experimentation on real-world datasets. In conclusion, this article has methodological implications for the stable assessment of students' knowledge and provides a reference for personalization and use of intelligence in the educational teaching process.

Purpose: Research shows that postsecondary students are largely unaware of the impact of algorithms on their everyday lives. Also, most noncomputer science students are not being taught about algorithms as part of the regular curriculum. This exploratory, qualitative study aims to explore subject-matter experts' insights and perceptions of the knowledge components, coping behaviors and pedagogical considerations to aid faculty in teaching algorithmic literacy to postsecondary students. Design/methodology/approach: Eleven semistructured interviews and one focus group were conducted with scholars and teachers of critical algorithm studies and related fields. A content analysis was manually performed on the transcripts using a mixture of deductive and inductive coding. Data analysis was aided by the coding software program Dedoose (2021) to determine frequency totals for occurrences of a code across all participants along with how many times specific participants mentioned a code. Then, findings were organized around the three themes of knowledge components, coping behaviors and pedagogy. Findings: The findings suggested a set of 10 knowledge components that would contribute to students' algorithmic literacy along with seven behaviors that students could use to help them better cope with algorithmic systems. A set of five teaching strategies also surfaced to help improve students' algorithmic literacy. Originality/value: This study contributes to improved pedagogy surrounding algorithmic literacy and validates existing multi-faceted conceptualizations and measurements of algorithmic literacy.

In this article, we delve into a hermeneutic process that analyzes the term "Computational Thinking" as it was constructed through Wing's series of iterations in conceptualization attempts (2006, 2008, 2011 and 2014). On the one hand, this brings us to analyze the relations and intersections between different process of thought (analytical, logical, mathematical, system, engineering, algorithmic) and the role of search for simplicity, generalization, and scalability in the layers of abstraction in Computational Thinking. On the other hand, we explore the roots and the discursive environment of authors that could contribute to this conceptualizing process in the years around its popularization following Jeannette Wing's founding article in 2006. We have also included in our analysis the points of convergence with Seymour Papert's work related to the computer-machine seen as an "object-to-think-with" helping to the construction of knowledge from an epistemological perspective to the computing-human. Even though a consensus on the definition of the concept has not yet been encountered, the analysis helps to highlight the solid reference points that address what is at the core of Computational thinking and what should be the framework of educational interventions and research that revolve around it.

With the popularity of online one-to-one tutoring, there are emerging concerns about the quality and effectiveness of this kind of tutoring. Although there are some evaluation methods available, they are heavily relied on manual coding by experts, which is too costly. Therefore, using machine learning to predict instruction quality automatically is an effective way to reduce human costs. Three classification methods are analyzed in this article: (1) random forest algorithm with human-engineered descriptive features; (2) long and short-term memory algorithm with acoustic features generated by open speech and music interpretation by large space extraction toolkit; and (3) convolutional neural network algorithm with Mel spectrogram of the audio. The results show that the three approaches can complete the prediction task well, with the second approach exhibiting the best accuracy. The importance of the features in these classification models is analyzed according to eXplainable Artificial Intelligence techniques (i.e., XAI) and statistical feature analysis methods. In this way, key indicators of high-quality tutoring are identified. This study demonstrated the usefulness of XAI techniques in understanding why some tutoring sessions are of good quality and others are not. The results can be potentially used to guide the improvement of online one-to-one tutoring in the future.

Students are the future of a nation. Personalizing student interests in higher education courses is one of the biggest challenges in higher education. Various AI and ML approaches have been used to study student behaviour. Existing AI and ML algorithms are used to identify features for various fields, such as behavioural analysis, economic analysis, image processing, and personalized medicine. However, there are major concerns about the interpretability and understandability of the decision made by a model. This is because most AI algorithms are black-box models. In this study, explain- able AI (XAI) aims to break the black box nature of an algorithm. In this study, XAI is used to identify engineering students' interests, and BRB and SP-LIME are used to explain which attributes are critical to their studies. We also used (PCA) for feature selection to identify the student cohort. Clustering the cohort helps to analyse the between influential features in terms of engineering discipline selection. The results show that there are some valuable factors that influence their study and, ultimately, the future of a nation.

Diagnostic uncertainty is prevalent throughout medicine and significantly impacts patient care, especially when it goes unrecognized. However, we lack a reliable clinical means of identifying uncertainty. This study evaluates the narrative discourse within clinical notes in the Electronic Health Record as a means of identifying diagnostic uncertainty. Recognizing that discourse producers use language "semi-automatically" (Partington et al. 2013), we hypothesized that clinicians include distinct indications of uncertainty in their written assessments, which could be elucidated by linguistic analysis. Using a cohort of patients prospectively identified as having an uncertain diagnosis (UD), we conducted a detailed corpus-assisted discourse analysis. The analysis revealed a set of linguistic indicators constitutive of diagnostic uncertainty including terms of modality, register-specific terms, and linguistically identifiable clinical behaviours. This dictionary of UD indicators was thoroughly tested, and its performance was compared with a matched-control dataset. Based on the findings, we built a machine learning classification algorithm with the ability to predict UD patient cohorts with 87.0% accuracy, effectively demonstrating the feasibility of using clinical discourse to classify patients and directly impact the clinical environment.

The application of machine learning (ML) has grown and is now used to enhance learning outcomes. In blended classroom settings, ML, emerging smartphones and wearable technologies are commonly used to improve teaching and learning. The combination of these advanced technologies and ML plays a crucial role in enhancing real-time feedback quality. However, there are abundant scopes of improvement and strong need for further careful investigations in this area. We propose an "ML-based intelligent real-time feedback system" to address current research challenges for blended classrooms. The proposed system provides real-time feedback to students and teachers. We build an Android application for our intelligent feedback interfaces. The user interfaces use students' academic performance prediction models with real-time states and dynamic feedback timings based on historic feedback statistics. In addition, the feedback scheduling algorithms, choices of peripheral devices for real-time feedback, and feedback modalities to optimize fatigue make our system interfaces intelligent and novel. The end users well-received the intelligent features and technology of the proposed system. Our empirical findings indicate that unique design elements, such as dynamic timing, choice of peripheral devices, and modalities of real-time feedback, are crucial in integrating the system with blended classes. The intelligent characteristics of the proposed system have been appreciated by a large proportion of the end-users (90.90% of teachers and 84.21% of students) for use in real-time blended classroom environments. The higher comparative system usability scale (SUS) scores with benchmarks show real promise of the system design.

Contribution: A problem-solving approach (PSA) model derived from major computational thinking (CT) concepts. This model can be utilized to formulate solutions for different algorithmic problems and translate them into effective active learning methods. Background: Different teaching approaches for programming are widely available; however, being able to formulate an algorithmic solution computationally and then transform it into code is essential for students. Research Questions: What are the effective teaching approaches for fostering the development of problem-solving and programming skills?, How do CT concepts contribute to the formulation of a PSA model for programming problems and its translation into an effective teaching method?, How can an effective teaching method that utilizes the PSA model be identified and distinguished from other approaches? Methodology: A preliminary study pointed out the difficulties experienced when teaching programming, inspiring the formulation of a PSA model that used CT concepts. An experimental study on problem-based and game-based programming workshops that utilized the PSA model through sorting algorithms was performed on experimental groups consisting of 30 students each. A syntax-based programming workshop consisting of 30 students was used as the control group. All the participants were recruited through a pretest that incorporated basic programming questions. The participants had to answer a posttest after the workshop. Findings: The results showed that the participants exhibited no significant difference between the pretest and posttest for the syntax-based learning (SBL). However, there is a significant difference between the pretest and posttest of both the problem-based learning (PBL) and the game-based learning (GBL) workshops. There was no significant difference for the pretest scores of all three workshops. The analysis of the posttest further confirmed that the experimental groups (PBL and GBL) exhibited significant difference in the scores compared to the control group. However, the posttest results did not differ significantly between the experimental groups (PBL and GBL).

Contribution: This research illuminates information entropy's efficacy as a pivotal educational tool in programming, enabling the precise quantification of algorithmic complexity and student abstraction levels for solving problems. This approach can provide students quantitative, comparative insights into the differences between optimal and student implemented solution, and allowing educators to offer targeted feedback, thereby optimizing the learning and abstraction processes in algorithm design through deliberate practice. Background: Abstraction is considered one of the most important skills in problem solving. Many studies in programming have shown that higher abstraction capability can significantly simplify problems, reduce program complexity and improve efficiency. However, it is difficult to develop criteria to measure the level of abstraction, and there is still a lack of relevant systematic research. Research Questions: 1) How can students' abstraction ability in programming be effectively measured? and 2) How to develop programming education and training methods based on the measurement of abstraction ability? Methodology: Forty-six grade 10 students participated in the experiment, divided into two groups for programming training using information-entropy-based assessment and traditional learning methods. Their level of computational thinking, algorithmic efficiency improvements, and test scores were used to measure performance and to analyze the effectiveness of the training methods. Findings: Through empirical research, this article finds that information-entropy-based assessment can reflect the differences in problem solving among students possessing varying capabilities. Information entropy can be crucial for evaluating and improving students' abstraction performance and algorithm efficiency.

Progress tests (PT) are a popular type of longitudinal assessment used for evaluating clinical knowledge retention and long-life learning in health professions education. Most PTs consist of multiple-choice questions (MCQs) whose development is costly and time-consuming. Automatic Item Generation (AIG) generates test items through algorithms, promising to ease this burden. However, it remains unclear how AIG-items behave in formative assessment (FA) modalities such as PTs compared to manually written items. The purpose of this study was to compare the quality and validity of AIG-items versus manually written items. Responses to 126 (23 automatically generated) dichotomously scored single best-answer five-option MCQs retrieved from the 2021 University of Minho PT of medicine were analyzed. Procedures based on item response theory (IRT), dimensionality testing, item fit, reliability, differential item functioning (DIF) and distractor analysis were used. Qualitative assessment was conducted through expert review. Validity evidence of AIG-items was assessed by using hierarchical linear modeling (HLM). The PT proved to be a viable tool for assessing medical students cognitive competencies. AIG-items were parallel to manually written-items, presenting similar indices of difficulty and information. The proportion of functional distractors for both AIG and manually written items was similar. Evidence of validity for AIG-items was found while showing higher levels of item quality. AIG-items functioned as intended and were appropriate for evaluating medical students at various levels of the knowledge spectrum.

Purpose: This paper aims to center the experiences of three cohorts (n = 40) of Black high school students who participated in a critical race technology course that exposed anti-blackness as the organizing logic and default setting of digital and artificially intelligent technology. This paper centers the voices, experiences and technological innovations of the students, and in doing so, introduces a new type of digital literacy: critical race algorithmic literacy. Design/methodology/approach: Data for this study include student interviews (called "talk backs"), journal reflections and final technology presentations. Findings: Broadly, the data suggests that critical race algorithmic literacies prepare Black students to critically read the algorithmic word (e.g. data, code, machine learning models, etc.) so that they can not only resist and survive, but also "rebuild" and "reimagine" the algorithmic world. Originality/value: While critical race media literacy draws upon critical race theory in education - a theorization of race, and a critique of white supremacy and multiculturalism in schools - critical race algorithmic literacy is rooted in critical race technology theory, which is a theorization of blackness as a technology and a critique of algorithmic anti-blackness as the organizing logic of schools and AI systems.

Purpose: Artificial intelligence (AI) has become increasingly important and influential in reading and writing. The influx of social media digital spaces, like TikTok, has also shifted the ways multimodal composition takes place alongside AI. This study aims to argue that within spaces like TikTok, human composers must attend to the ways they write for, with and against the AI-powered algorithm. Design/methodology/approach: Data collection was drawn from a larger study on #BookTok (the TikTok subcommunity for readers) that included semi-structured interviews including watching and reflecting on a TikTok they created. The authors grounded this study in critical posthumanist literacies to analyze and open code five #BookTok content creators' interview transcripts. Using axial coding, authors collaboratively determined three overarching and entangled themes: writing for, with and against. Findings: Findings highlight the nuanced ways #BookTokers consider the AI algorithm in their compositional choices, namely, in the ways how they want to disseminate their videos to a larger audience or more niche-focused community. Throughout the interviews, participants revealed how the AI algorithm was situated differently as both audience member, co-author and censor. Originality/value: This study is grounded in critical posthumanist literacies and explores composition as a joint accomplishment between humans and machines. The authors argued that it is necessary to expand our human-centered notions of what it means to write for an audience, to co-author and to resist censorship or gatekeeping.

As algorithmic decision making is increasingly deployed in every walk of life, many researchers have raised concerns about fairness-related bias from such algorithms. But there is little research on harnessing psychometric methods to uncover potential discriminatory bias inside decision-making algorithms. The main goal of this article is to propose a new framework for algorithmic fairness based on "differential item functioning" (DIF), which has been commonly used to measure item fairness in psychometrics. Our fairness notion, which we call "differential algorithmic functioning" (DAF), is defined based on three pieces of information: a decision variable, a "fair" variable, and a protected variable such as race or gender. Under the DAF framework, an algorithm can exhibit uniform DAF, nonuniform DAF, or neither (i.e., non-DAF). For detecting DAF, we provide modifications of well-established DIF methods: Mantel-Haenszel test, logistic regression, and residual-based DIF. We demonstrate our framework through a real dataset concerning decision-making algorithms for grade retention in K-12 education in the United States.

Eye gaze tracking has recently become indispensable for domains like virtual reality, augmented reality, human-computer interaction and advertisement. The commercial eye gaze tracking equipment is too expensive to be used by the masses. In this manuscript, a non-invasive, low-resolution ordinary camera-based system has been proposed for tracking eye gaze. The proposed system attained an accuracy of 84% without requiring any costly equipment or infrared light like other commercial tools. Eye gaze fixations data of 37 participants were recorded on a webpage and gaze patterns were plotted using heat maps. Further, a case analysis was done to validate the performance of the proposed system, where the user interface has been controlled using eye movements.

In this paper, we develop a mixed stochastic approximation expectation-maximization (MSAEM) algorithm coupled with a Gibbs sampler to compute the marginalized maximum a posteriori estimate (MMAPE) of a confirmatory multidimensional four-parameter normal ogive (M4PNO) model. The proposed MSAEM algorithm not only has the computational advantages of the stochastic approximation expectation-maximization (SAEM) algorithm for multidimensional data, but it also alleviates the potential instability caused by label-switching, and then improved the estimation accuracy. Simulation studies are conducted to illustrate the good performance of the proposed MSAEM method, where MSAEM consistently performs better than SAEM and some other existing methods in multidimensional item response theory. Moreover, the proposed method is applied to a real data set from the 2018 Programme for International Student Assessment (PISA) to demonstrate the usefulness of the 4PNO model as well as MSAEM in practice.

Personalized educational interventions have been shown to facilitate successful and inclusive statistics, mathematics, and data science (SMDS) in higher education through timely and targeted reduction of heterogeneous training disparities caused by years of cumulative, structural challenges in contemporary educational systems. However, the burden on the institutions and instructors to provide personalized training resources to large groups of students is also formidable, and often unsustainable. We present Individualized Pathways and Resources to Adaptive Control Theory-Inspired Scientific Education (iPRACTISE), a free, publicly available web app that serves as a tool to facilitate personalized trainings on SMDS and related topics through provision of personalized training recommendations as informed by computerized assessments and individuals' training preferences. We describe the resources available in iPRACTISE, and some proof-of-concept evaluation results from deploying iPRACTISE to supplement in-person and online classroom teaching in real-life settings. Strengths, practical difficulties, and potentials for future applications of iPRACTISE to crowdsource and sustain personalized SMDS education are discussed.

Support vector classifiers are one of the most popular linear classification techniques for binary classification. Different from some commonly seen model fitting criteria in statistics, such as the ordinary least squares criterion and the maximum likelihood method, its algorithm depends on an optimization problem under constraints, which is unconventional to many students in a second or third course in statistics or data science. As a result, this topic is often not as intuitive to students as some of the more traditional statistical modeling tools. In order to facilitate students' mastery of the topic and promote active learning, we developed some in-class activities and their accompanying Shiny applications for teaching support vector classifiers. The designed course materials aim at engaging students through group work and solidifying students' understanding of the algorithm via hands-on explorations. The Shiny applications offer interactive demonstration of the changes of the components of a support vector classifier when altering its determining parameters. With the goal of benefiting the broader statistics and data science education community, we have made the developed Shiny applications publicly available. In addition, a detailed in-class activity worksheet and a real data example are also provided in the online supplementary materials.

This study investigated the effect of video lecture types on the performance of students in computational problem-solving practices. A total of 19 university students participated in the computational problem-solving practices that mostly required declarative knowledge, and 22 university students participated in the computational problem-solving practices that mostly required procedural knowledge. The practices were implemented in the Algorithm and Programming course and the Computer Programming II course. Three video lecture types (instructor-whiteboard, instructor voice-handbook, instructor-slides) were used in both courses. The one-way repeated measures ANOVA test was employed to determine if there was a significant difference between the problem-solving performances of the students based on the video lecture type. In the Algorithm and Programming course that required mostly declarative knowledge, the problem-solving scores of the students were significantly higher after the instructor voice-handbook video practice than those after the instructor-whiteboard video practice. On the other hand, in the Computer Programming II course that required mostly procedural knowledge, the problem-solving scores of the students were significantly higher after the instructor-whiteboard video practice than those after the instructor voice-handbook video practice. The students showed higher performance in the video lecture types they preferred in both courses. The students listed the factors that affect their video preferences as: (a) the effect of the presence of an instructor in the video lecture on their attention, (b) the efficiency of the video lecture in examining many and various examples in a limited time, (c) the opportunity provided by the video lecture to revise the content and procedure, and (d) the efficient presentation of the knowledge. It is recommended that an instructor should be present in the video that includes mostly procedural knowledge, while there is no need for an instructor in the video that includes mostly declarative knowledge regarding computational problem-solving activities.

With the advent of technological advancement in learning, such as context-awareness, ubiquity and personalisation, various innovations in teaching and learning have led to improved learning. This research paper aims to develop a system that supports personalised learning through adaptive content, adaptive learning path and context awareness to meet individual learner's requirements and promote the effectiveness and performance of the learning process. Furthermore, the study used an experimental control group to ascertain the significance of learners' learning styles and preferences in improving learning performance. The model considers different modules such as personalisation, location support, learning plan and context awareness while relying on the proper classification of learners' learning styles. The system was experimented using 127 learners of a Computer Science Course. The result obtained using 10-fold cross-validation on various algorithms shows better performance in ROC, Kappa and Accuracy, which were all used to evaluate the quality of classification outcome and compared with learners' performances in assessments. During classification, Naïve Bayes Model has the highest average ROC value of 0.979, TP Rate of 0.951 and FP Rate of 0.045. The results obtained show an improvement in learning.

The purpose of this study is to introduce and evaluate a method for generating reading comprehension items using template-based automatic item generation. To begin, we describe a new model for generating reading comprehension items called the text analysis cognitive model assessing inferential skills across different reading passages. Next, the text analysis cognitive model is used to generate reading comprehension items where examinees are required to read a passage and identify the irrelevant sentence. The sentences for the generated passages were created using OpenAI GPT-3.5. Finally, the quality of the generated items was evaluated. The generated items were reviewed by three subject-matter experts. The generated items were also administered to a sample of 1,607 Grade-8 students. The correct options for the generated items produced a similar level of difficulty and yielded strong discrimination power while the incorrect options served as effective distractors. Implications of augmented intelligence for item development are discussed.

Founded in 2006, Zillow established itself as the leading online real estate marketplace. In 2018, Zillow launched Zillow Offers, a new business that purchased and sold homes. Zillow Offers provided home sellers with a faster purchase process than traditional realtors by gathering data from sellers online and making offers immediately, a process known as "iBuying" (i.e., "Instant Buying"). Though new to iBuying, Zillow quickly established a goal of generating $20 billion in annual revenue within three-to-five years. Zillow believed that its artificial intelligence/machine learning (AI/ML) platform for predicting home values, aka "Zestimate," could be a competitive advantage in the iBuying marketplace. However, after losing $421 million in its iBuying business during the third quarter of 2021, the company closed this once-promising business unit rather than risk further losses. CEO Rich Barton asserted that the AI's inability to accurately predict home prices caused the failure of its iBuying business. This case study examines the trajectory of Zillow Offers and discusses several factors that contributed to its demise. After exploring the challenges of its home-price prediction algorithms within iBuying, we argue that the failure of Zillow Offers extends beyond the limitations of its initial AI/ML system. Zillow Offers' focus on hypergrowth over profitability led to operational changes that failed to balance estimated price predictions and operational purchase price decisions. Through this analysis, we raise important questions for students and practitioners about the appropriate and effective use of data-driven AI/ML models for operational decision making.

The default risk, one of the main risk factors for bonds, should be measured and reflected in the bond yield. Particularly, in the case of financial companies that treat bonds as a major product, failure to properly identify and filter customers' workout status adversely affects returns. This study proposes a two-stage classification algorithm for workout prediction based on the history data of individual customers such as transaction details of financial companies secured after loans, which is collected over 10 years. The first stage is to rank variables that are closely related to the workout application based on feature selection. In the second step, the first to nth cumulative variables input to each machine learning method generate n candidate classifiers, respectively. Among the total candidates, the model with the highest classification accuracy was selected as the optimal one, which is the Gradient Boost combined with F-score-based feature selection.

It is meaningful to create a comprehensive score to extract information from mass continuous data when they measure the same latent concept. Therefore, this study adopts the logic of psychometrics to conduct scales on continuous data under the Rasch models. This study also explores the effect of different data discretization methods on scale development by using financial profitability ratios as a demonstration. Results show that retaining more categories can benefit Rasch modeling because it can better inform the models. The dynamic clustering algorithm, k-median is a better method for extracting characteristic patterns of the continuous data and preparing the data for the Rasch model. This study illustrates that there is no one-way good discretization method for continuous data under the Rasch models. It is more reasonable to use the traditional algorithms if each continuous data variable has target benchmark(s), whereas the k-median clustering algorithm achieves good modeling results when benchmark information is lacking.

Computer science development, especially machine learning, is a thriving innovation essential for education. It makes the process of teaching and learning more accessible and manageable and also promotes equality. The positive influence of machine learning can also be felt in Islamic studies, particularly in Hadith studies. This literature review highlights the role of machine learning in managing research regarding Hadith studies that have been published and categorizing it by their research topics, language & corpus, and the machine-learning algorithms. This article review has been conducted on 48 previously published hadith study journals. Then, we summarize existing trends, including trending topics, common language & corpus, and general algorithms often used in previous hadith-related reviews. This article aims to give new insight to help the broad community of researchers interested in these narrations to create fresh and further research with the uncommon topic, language & corpus, and algorithms. Furthermore, this article is also expected to contribute to academics and practitioners as a guide for conducting future research on the application of computer science in Hadith studies. We conclude that the most frequently discussed topic is Hadith Classification at 33.33%, the most widely used language is Arabic at 43.75%, and the most commonly used algorithm is SVM at 12.5%. In addition, the dataset mainly used is a public dataset by Al-Bukhari at 30.53%.

Predicting student performance using educational data is a significant area of machine learning research. However, class imbalance in datasets and the challenge of developing interpretable models can hinder accuracy. This study compares different variations of the Synthetic Minority Oversampling Technique (SMOTE) combined with classification algorithms to create prediction models. The results show that SMOTE with Edited Nearest Neighbors is superior, and the balanced random forest classifier performs better when using SMOTE-ENN, achieving 96% accuracy, precision, and F-value. Smote also has faster execution time. For model interpretability, combining Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP) provides deeper insights. LIME is suitable for single-prediction interpretation, while SHAP is better for overall model interpretation. This research offers guidelines to mitigate data imbalance and improve fairness in education through data-driven innovations like early warning systems. It also educates academics on explainability approaches to facilitate wider use of machine learning methods.

Machine learning models were constructed to predict student performance in an introductory mechanics class at a large land-grant university in the United States using data from 2061 students. Students were classified as either being at risk of failing the course (earning a D or F) or not at risk (earning an A, B, or C). The models focused on variables available in the first few weeks of the class which could potentially allow for early interventions to help at-risk students. Multiple types of variables were used in the model: in-class variables (average homework and clicker quiz scores), institutional variables [college grade point average (GPA)], and noncognitive variables (self-efficacy). The substantial imbalance between the pass and fail rates of the course, with only about 10% of students failing, required modification to the machine learning algorithms. Decision threshold tuning and upsampling were successful in improving performance for at-risk students. Logistic regression combined with a decision threshold tuned to maximize balanced accuracy yielded the strongest classifier, with a DF accuracy of 83% and an ABC accuracy of 81%. Measures of variable importance involving changes in balanced accuracy identified homework grades, clicker grades, college GPA, and the fraction of college classes successfully completed as the most important variables in predicting success in introductory physics. Noncognitive variables added little predictive power to the models. Classification models with performance near the best-performing models using the full set of variables could be constructed with very few variables (homework average, clicker scores, and college GPA) using straightforward to implement algorithms, suggesting the application of these technologies may be fairly easy to include in many physics classes.

This article intends to empower and equip art educators to artistically address the functioning of facial detection algorithms through critical race theory (CRT). By highlighting how biometric data, a specific form of data that measures the physical qualities of individuals, is used in common social media facial detection algorithms like Snapchat, art students are encouraged to use principles of value and facial proportions to interact with and explore these algorithms in a material way. Examples of algorithms systematically perpetuating racism are shared to highlight the real-world significance of data and algorithms. Artistic responses to biometrics and artificial intelligence (AI) are provided to inspire student artistic research, along with teacher examples. Finally, the conclusion urges art educators to address the functioning of algorithms and their potential issues through the arts.

In-class group work activities are found to promote the interpersonal skills of learners. To support the teachers in facilitating such activities, we designed a learning analytics-enhanced technology framework, Group Learning Orchestration Based on Evidence (GLOBE) using data-driven approaches. In this study, we implemented the algorithmic group formation and group work evaluation systems in a Japanese junior high school context. Data from a series of 12 collaborative learning activities were used to validate the difference in the measured heterogeneity of the formed homogeneous and heterogeneous groups compared to random grouping. Further, the peer rating and self-perception of the group work were compared for comparative reading and idea exchange tasks. We found algorithmically formed groups, considering the learner model data, either heterogeneously or homogeneously performed better than random grouping. Specifically, students in groups created by the homogeneous algorithm received higher peer ratings and more positive self-perception of group work in the idea exchange group tasks. We did not find significant differences in the comparative reading tasks. Along with the empirical findings, this work presents a paradigm of continuous data-driven group learning support by incorporating the peer and teacher evaluation scores as an input to the subsequent algorithmic grouping.

Today, in every academic institution as well as the university system assessing students' performance, identifying the uniqueness of each student and finding solutions to performance problems have become challenging issues. The main purpose of the study is to predict how student performance changes as a result of their behaviours, hobbies, extracurricular activities and different university activities. This study collected data from graduates via the online and supervised machine learning algorithms used to solve the problem. After pre-processing data, classification algorithms were applied, namely Random Forest, Multi-Layer Perceptron, Support Vector Machine, Naïve Bayes and Decision Tree. The results show that the Multi-Layer Perceptron is the best algorithm considering the highest accuracy and lowest error values. An ensemble learning algorithm was then applied by combining those five algorithms. The best results were obtained using it, and according to the final results, ensemble learning increases the accuracy rather than each classifier.

The world's move to a global economy has an impact on the high rate of student academic failure. Higher education, as the affected party, is considered crucial in reducing student academic failure. This study aims to construct a prediction (predictive model) that can forecast students' time to graduation in developing countries such as Indonesia, as well as the essential factors (attributes) that can explain it. This research used a data mining method. The data set used in this study is from an Indonesian university and contains demographic and academic records of 132,734 students. Demographic data (age, gender, marital status, employment, region, and minimum wage) and academic (i.e., grade point average (GPA)) were utilized as predictors of students' time to graduation. The findings of this study show that (1) the prediction model using the random forest and neural networks algorithms has the highest classification accuracy (CA), and area under the curve (AUC) value in predicting students' time to graduation (CA: 76% and AUC: 79%) compared to other models such as logistic regression, Naïve Bayes, and k-nearest neighbor; and (2) the most critical variable in predicting students' time to graduation along with six other important variables is the student's GPA.

In this paper, to handle the problem of the quick evolution of cyber-security attacks, we developed the iMonsters board game and proposed the attack and defense knowledge self-evolving algorithm. Three versions of the iMonsters were launched in 2013, 2017, and 2019, respectively. Accordingly, the cyber-security ontology can be refined by the ontology fusion-or-splitting procedure for the newly collected cyber-security incidents, as well as the roles and rules of the iMonsters can be refined by the gaming portfolio mining procedure for the collected portfolio. Furthermore, we conducted game-based learning (GBL), a quasi-experiment of pre/post-testing, and concept map testing using the iMonsters board game in a children's summer camp. The experimental results indicate that the students can acquire up-to-date cyber-security knowledge with the iMonsters better than students who learn in a traditional classroom setting, and the students' satisfaction with acquiring cyber-security knowledge with the instructional design of the iMonsters is better compared with learning in a traditional classroom setting. Satisfaction with the new version has continuously increased. Besides, the results of the in-depth interviews show that the new version was easier to learn. Thus, we may conclude that using the self-evolving iMonsters can improve learning effectiveness and participation in GBL.

A key commitment of computer-supported collaborative learning research is to study how people learn in collaborative settings to guide development of methods for capture and design for learning. Computer-supported collaborative learning research has a tradition of studying how the physical world plays a part in collaborative learning. Within the field, a material turn is emerging that considers how digital and tangible technologies actively contribute to collaborative learning processes. Studying how tangible materials produce collaborative learning visibly and algorithmically is particularly important at a time when advanced algorithms are integrated into educational contexts in ways that are not always transparent. However, the needed methodologies for capturing how non-human agents take part in collaborative learning remains underdeveloped. The present study builds on current CSCL research that investigates materials in collaborative learning and introduces posthumanist perspectives with the aim to decenter humans methodologically and to probe empirically whether and how these perspectives contribute to empirical understanding of collaborative learning processes. Taking fiber crafts (e.g., weaving and fabric manipulation) as a context for computational learning, the present study conducted a posthumanist "analysis of differences" among human and non-human participants in collaboration using video data to investigate how middle school youths and fiber craft components performed algorithms over time. The findings show how both youths and craft materials actively contributed to the performance of algorithms. In weaving, algorithms became repeated youth-material movements one dimension at a time. In fabric manipulation, algorithms became a repeated confluence of component parts. Decentering humans through an analysis of differences among human and non-human introduced human-material collaboration as a productive unit of analysis for understanding how materials and people together contribute to producing what can be recognized as computational performance. The findings of this research contribute to ongoing conversations in CSCL research on how computational materials can be considered in collaborative learning and present a new approach to capture collaborative learning as physical expansion over time. The study has implications for future research on capturing collaborative computational learning and designing physical computational learning opportunities that show technology as evolving.

Recently, biological aging has been quantified in DNA-methylation samples of older adults and applied as so-called "methylation profile scores" (MPSs) in separate target samples, including samples of children. This nascent research indicates that (1) biological aging can be quantified early in the life course, decades before the onset of aging-related disease, (2) is affected by common environmental predictors of childhood development, and (3) shows overlap with "developmental processes" (e.g., puberty). Because the MPSs were computed using algorithms developed in adults, these studies indicate a molecular link between childhood environments, development, and adult biological aging. Yet, if MPSs can be used to connect development and aging, previous research has only traveled one way, deriving MPSs developed in adults and applying them to samples of children. Researchers have not yet quantified epigenetic measures that reflect the pace of child development, and tested whether resulting MPSs are associated with physical and psychological aging. In this perspective I posit that combining measures of biological aging with new quantifications of child development has the power to address fundamental questions about life span: How are development and experience in childhood related to biological aging in adulthood? And what is aging?

Generalized structured component analysis (GSCA) is a multivariate method for specifying and examining interrelationships between observed variables and components. Despite its data-analytic flexibility honed over the decade, GSCA always defines every component as a linear function of observed variables, which can be less optimal when observed variables for a component are nonlinearly related, often reducing the component's predictive power. To address this issue, we combine deep learning and GSCA into a single framework to allow a component to be a nonlinear function of observed variables without specifying the exact functional form in advance. This new method, termed deep learning generalized structured component analysis (DL-GSCA), aims to maximize the predictive power of components while their directed or undirected network remains interpretable. Our real and simulated data analyses show that DL-GSCA produces components with greater predictive power than those from GSCA in the presence of nonlinear associations between observed variables per component.

In confirmatory factor analysis (CFA), model parameters are usually estimated by iteratively minimizing the Maximum Likelihood (ML) fit function. In optimal circumstances, the ML estimator yields the desirable statistical properties of asymptotic unbiasedness, efficiency, normality, and consistency. In practice, however, real-life data tend to be far from optimal, making the algorithm prone to convergence failure, inadmissible solutions, and bias. In this study, we revisited some old, yet largely neglected non-iterative alternatives and compared their performance to more recently proposed procedures in an extensive simulation study. We conclude that closed-form expressions may serve as viable alternatives for ML, with the Multiple Group Method -- the oldest method under consideration -- showing favorable results across all settings.

The human face is commonly used for identity verification. While this task was once exclusively performed by humans, technological advancements have seen automated facial recognition systems (AFRS) integrated into many identification scenarios. Although many state-of-the-art AFRS are exceptionally accurate, they often require human oversight or involvement, such that a human operator actions the final decision. Previously, we have shown that on average, humans assisted by a simulated AFRS (sAFRS) failed to reach the level of accuracy achieved by the same sAFRS alone, due to overturning the system's correct decisions and/or failing to correct sAFRS errors. The aim of the current study was to investigate whether participants' trust in automation was related to their performance on a one-to-one face matching task when assisted by a sAFRS. Participants (n = 160) completed a standard face matching task in two phases: an unassisted baseline phase, and an assisted phase where they were shown the identification decision (95% accurate) made by a sAFRS prior to submitting their own decision. While most participants improved with sAFRS assistance, those with greater relative trust in automation achieved larger gains in performance. However, the average aided performance of participants still failed to reach that of the sAFRS alone, regardless of trust status. Nonetheless, further analysis revealed a small sample of participants who achieved 100% accuracy when aided by the sAFRS. Our results speak to the importance of considering individual differences when selecting employees for roles requiring human-algorithm interaction, including identity verification tasks that incorporate facial recognition technologies.

Although the generalized deterministic inputs, noisy "and" gate model (G-DINA; de la Torre, 2011) is a general cognitive diagnosis model (CDM), it does not account for the heterogeneity that is rooted from the existing latent groups in the population of examinees. To address this, this study proposes the mixture G-DINA model, a CDM that incorporates the G-DINA model within the finite mixture modeling framework. An expectation--maximization algorithm is developed to estimate the mixture G-DINA model. To determine the viability of the proposed model, an extensive simulation study is conducted to examine the parameter recovery performance, model fit, and correct classification rates. Responses to a reading comprehension assessment were analyzed to further demonstrate the capability of the proposed model.

Email filters classify new messages as either spam or not spam based on word frequency, syntax, and metadata. A "classifier" is an algorithm that maps input data into categories based on distinguishing characteristics, or "features." Features can be raw data or attributes derived from that data. "Feature engineering" is the process of identifying or extracting features that can be used to capture relevant patterns in data. The project that the author introduces in this article uses feature engineering with a dataset called the Modified National Institute of Standards and Technology dataset (MNIST). MNIST is a modified version of a NIST dataset that contains 70,000 grayscale images of handwritten digits, and the goal with this dataset is to build a classifier that can accurately recognize digits from those images. Students use feature engineering to build a classifier that can accurately recognize digits from images.

The questionnaire method has always been an important research method in psychology. The increasing prevalence of multidimensional trait measures in psychological research has led researchers to use longer questionnaires. However, questionnaires that are too long will inevitably reduce the quality of the completed questionnaires and the efficiency of collection. Computer adaptive testing (CAT) can be used to reduce the test length while preserving the measurement accuracy. However, it is more often used in aptitude testing and involves a large number of parametric assumptions. Applying CAT to psychological questionnaires often requires question-specific model design and preexperimentation. The present article proposes a nonparametric and item response theory (IRT)-independent CAT algorithm. The new algorithm is simple and highly generalizable. It can be quickly used in a variety of questionnaires and tests without being limited by theoretical assumptions in different research areas. Simulation and empirical studies were conducted to demonstrate the validity of the new algorithm in aptitude tests and personality measures.

Educational process mining aims (EPM) to help teachers understand the overall learning process of their students. Although deep learning models have shown promising results in many domains, the event log dataset in many online courses may not be large enough for deep learning models to approximate the probability distribution of students' learning sequence due to a lack of participants. This study proposes a deep learning framework to help uncover the learning progression of learners. It aims to produce a graphical representation of the overall educational process from event logs. Our framework adopts the Smith-Waterman algorithm from the bioinformatics field to evaluate general learning sequences generated from deep learning models. Using our framework, we compare the performance of a deep learning model with the modified cross-attention layer and a model without modification and find that the modified model outperforms the other. The contribution of this framework is that it enables the use of neural architecture search techniques to uncover students' general learning sequence irrespective of the dataset's size. The framework also helps educators identify education materials that present as learning bottlenecks, enabling them to improve the materials and their respective layout order, thus facilitating student learning.