Search

Abstract Educational recommender systems have been supporting personalized learning in various ways. However, less discussion is conducted about whether and how to personalize the strategies to generate recommendations based on student differences. In this study, we aim at investigating how students judge recommendations based on different strategies, and how these judgments relate to student characteristics. We conducted a large-scale questionnaire survey to measure students’ Big-Five personality traits, confidence in the subjects, and their judgments on six types of recommendations. The answers collected from 735 high school students in Japan indicate that students had different judgments across different recommendation strategies, but similarly for English and mathematics. Furthermore, the correlations between student characteristics and their judgments on recommendations were stronger if the subject to learn was inconsistent with the subject they preferred. The results provide insights on how to design educational recommendations that not only cater to students’ traits, but also help foster and enhance their traits for better learning.

Abstract In the age of artificial intelligence (AI), trust in AI systems is becoming more important. Explainable recommenders, which explain why an item is recommended, have recently been proposed in the field of learning technology to improve transparency, persuasiveness, and trustworthiness. However, the methods for generating explanations are limited and do not consider the learner’s cognitive perceptions or personality. This study draws inspiration from tailored intervention research in public health and investigates the effectiveness of personality-based tailored explanations by implementing them for the recommended quizzes in an explainable recommender system. High school students (n = 217) were clustered into three distinct profiles labeled Diligent (n = 77), Fearful (n = 72), and Agreeable (n = 68), based on the Big Five personality traits. The students were divided into a tailored intervention group (n = 106) and a control group (n = 111). In the tailored intervention group, personalized explanations for recommended quizzes were provided based on student profiles, with explanations based on quiz characteristics. In the control group, only non-personalized explanations based on quiz characteristics were provided. An 18-day A/B experiment showed that the tailored intervention group had significantly higher recommendation usage than the control group. These results suggest that personality-based tailored explanations with a recommender approach are effective for e-learning engagement and imply improved trustworthiness of AI learning systems.

Data-driven platforms with rich data and learning analytics applications provide immense opportunities to support collaborative learning such as algorithmic group formation systems based on learning logs. However, teachers can still get overwhelmed since they have to manually set the parameters to create groups and it takes time to understand the meaning of each indicator. Therefore, it is imperative to explore predictive indicators for algorithmic group formation to release teachers from the dilemma with explainable group formation indicators and recommended settings based on group work purposes. Employing learning logs of group work from a reading-based university course, this study examines how learner indicators from different dimensions before the group work connect to the subsequent group work processes and consequences attributes through correlation analysis. Results find that the reading engagement and previous peer ratings can reveal individual achievement of the group work, and a homogeneous grouping strategy based on reading annotations and previous group work experience can predict desirable group performance for this learning context. In addition, it also proposes the potential of automatic group formation with recommended parameter settings that leverage the results of predictive indicators.

Abstract In recent years, smart learning environments have become central to modern education and support students and instructors through tools based on prediction and recommendation models. These methods often use learning material metadata, such as the knowledge contained in an exercise which is usually labeled by domain experts and is costly and difficult to scale. It recognizes that automated labeling eases the workload on experts, as seen in previous studies using automatic classification algorithms for research papers and Japanese mathematical exercises. However, these studies didn’t delve into fine-grained labeling. In addition to that, as the use of materials in the system becomes more widespread, paper materials are transformed into PDF formats, which can lead to incomplete extraction. However, there is less emphasis on labeling incomplete mathematical sentences to tackle this problem in the previous research. This study aims to achieve precise automated classification even from incomplete text inputs. To tackle these challenges, we propose a mathematical exercise labeling algorithm that can handle detailed labels, even for incomplete sentences, using word n-grams, compared to the state-of-the-art word embedding method. The results of the experiment show that mono-gram features with Random Forest models achieved the best performance with a macro F-measure of 92.50%, 61.28% for 24-class labeling and 297-class labeling tasks, respectively. The contribution of this research is showing that the proposed method based on traditional simple n-grams has the ability to find context-independent similarities in incomplete sentences and outperforms state-of-the-art word embedding methods in specific tasks like classifying short and incomplete texts.

Abstract Digitized learning materials are a core part of modern education, and analysis of the use can offer insight into the learning behavior of high and low performing students. The topic of predicting student characteristics has gained a lot of attention in recent years, with applications ranging from affect to performance and at-risk student prediction. In this paper, we examine students reading behavior using a digital textbook system while taking an open-book test from the perspective of engagement and performance to identify the strategies that are used. We create models to predict the performance and engagement of learners before the start of the assessment and extract reading behavior characteristics employed before and after the start of the assessment in a higher education setting. It was found that strategies, such as: revising and previewing are indicators of how a learner will perform in an open ebook assessment. Low performing students take advantage of the open ebook policy of the assessment and employ a strategy of searching for information during the assessment. Also compared to performance, the prediction of overall engagement has a higher accuracy, and therefore could be more appropriate for identifying intervention candidates as an early-warning intervention system.

While data privacy is a key aspect of Learning Analytics, it often creates difficulty when promoting research into underexplored contexts as it limits data sharing. To overcome this problem, the generation of synthetic data has been proposed and discussed within the LA community. However, there has been little work that has explored the use of synthetic data in real-world situations. This research examines the effectiveness of using synthetic data for training academic performance prediction models, and the challenges and limitations of using the proposed data sharing method. To evaluate the effectiveness of the method, we generate synthetic data from a private dataset, and distribute it to the participants of a data challenge to train prediction models. Participants submitted their models as docker containers for evaluation and ranking on holdout synthetic data. A post-hoc analysis was conducted on the top 10 participant’s models by comparing the evaluation of their performance on synthetic and private validation datasets. Several models trained on synthetic data were found to perform significantly poorer when applied to the non-synthetic private dataset. The main contribution of this research is to understand the challenges and limitations of applying predictive models trained on synthetic data in real-world situations. Due to these challenges, the paper recommends model designs that can inform future successful adoption of synthetic data in real-world educational data systems.

The advent of blockchain technology over the last decade has led to the development of multiple use-cases of decentralization in various fields including education. This paper presents a unique bibliometric and qualitative analysis of the blockchain in education with novel contributions on temporal development, emerging themes and practical case studies on adoption and integration with existing educational technologies. We focus on identifying the major actors in the space, demographic participation and adoption, current hot topics, grey areas, and potential areas for innovation. Our analysis shows that while the blockchain has been around for about 13 years, blockchain in education only became prominent 5 years ago. This research also reveals that most of the efforts have been focused on reporting and verifying academic certificates and transcripts: only very few research focused on reporting and connecting in-depth academic records such as learning behaviour logs, learning contents and assessment data. This calls for concern as current education blockchain systems do not consider interoperability at the blockchain level and the heterogeneous nature in which institutes create and consume academic data. Finally, we present discussions on the implications of our findings, potential solutions and aspects of education blockchain research that can help to improve educational outcomes for various stakeholders.

In the realm of mathematics education, self-explanation stands as a crucial learning mechanism, allowing learners to articulate their comprehension of intricate mathematical concepts and strategies. As digital learning platforms grow in prominence, there are mounting opportunities to collect and utilize mathematical self-explanations. However, these opportunities are met with challenges in automated evaluation. Automatic scoring of mathematical self-explanations is crucial for preprocessing tasks, including the categorization of learner responses, identification of common misconceptions, and the creation of tailored feedback and model solutions. Nevertheless, this task is hindered by the dearth of ample sample sets. Our research introduces a semi-supervised technique using the large language model (LLM), specifically its Japanese variant, to enrich datasets for the automated scoring of mathematical self-explanations. We rigorously evaluated the quality of self-explanations across five datasets, ranging from human-evaluated originals to ones devoid of original content. Our results show that combining LLM-based explanations with mathematical material significantly improves the model’s accuracy. Interestingly, there is an optimal limit to how many synthetic self-explanation data can benefit the system. Exceeding this limit does not further improve outcomes. This study thus highlights the need for careful consideration when integrating synthetic data into solutions, especially within the mathematics discipline.

Abstract This study investigates learner’s reading behaviors in a critical reading task in humanities course using learning analytics techniques. A Critical Analysis of Literature and Cinema course was selected as a context. The course activities evolved over 10 years, and for this instance, some face-to-face classroom critical reading activities were migrated to online mode by using BookRoll, a learning analytics enhanced eBook platform. Students (n=22 out of the 50 registered) accessed Hayavadana, an Indian play uploaded on BookRoll, and attempted to identify performative elements and cultural references in the text and highlight them. In this study, we analyze learner’s reading logs gathered in the learning record store linked to BookRoll during that activity. We extend our previous work where we identify four online reading profiles: effortful, strategic, wanderers, and check-out, based on learner’s clickstream interactions and time spent with the content. We validate the profiles with qualitative interview data collected from the learners and illustrate the quantified learning behaviors of each of those profiles based on an engagement metric. Our work aims to initiate further discussion related to the application of learning analytics in humanities courses both to probe into the learning behaviors of the students and thereby enhance the experiences with the use of interactive learning environments and data-driven services.

This paper explores co-design in Japanese education for deploying data-driven educational technology and practice. Although there is a growing emphasis on data to inform educational decision-making and personalize learning experiences, challenges such as data interoperability and inconsistency with teaching goals prevent practitioners from participating. Co-design, characterized by involving various stakeholders, is instrumental in addressing the evolving needs of technology deployment. Japan's educational context aligns with co-design implementation, with a learning and evidence analytics infrastructure facilitating data collection and analysis. From the Japanese co-design practice of educational technologies, the paper highlights a 6-phase co-design framework: motivate, pilot, implement, refine, evaluate, and maintain. The practices focus on data-driven learning strategies, technology interventions, and across-context dashboards, covering assorted learning contexts in Japan. By advocating for a co-design culture and data-driven approaches to enhance education in Japan, we offer insights for education practitioners, policymakers, researchers, and industry developers.

This article originated from a working group on "Learning beyond formal schooling through human--computer--human interaction (HCHI)" convened at the UNESCO EDUSummIT 2023 in Kyoto (Japan). A polylogue approach was adopted by engaging eight co-authors whose diverse perspectives culminated in propositions that addressed the pivotal question: How should the connections between formal, non-formal, and informal learning be considered in a digitally mediated world? Formal learning is typically structured, organized and chronologically arranged institutional learning, whereas informal learning is associated with everyday learning across contexts throughout one's life, and non-formal learning is a hybrid of these forms of learning. Considering the growing prominence of informal and non-formal learning in a digitally mediated world, the evolving learning ecosystem calls for a recalibration of the emphasis on formal learning. In this regard, HCHI has the potential to mediate human--human interactions, thereby bridging formal and informal learning. Our articulated position is to preserve the distinct boundaries and inherent complexities of each type of learning while creating opportunities or 'bridges' to authentically draw on the processes of each through meaningful actions. The polylogue yielded three propositions to bridge the connections between formal, non-formal and informal learning spaces in a digitally mediated world: (a) formal education institutions can establish strategic alliances and collaborations with learning organizations beyond the formal educational system, (b) the creation of digital learning communities within formal and non-formal spaces exemplify a paradigm of operation in an unregulated, student-centric cohesive space where the educator is a co-participant, (c) the recognition of the value of informal learning experiences by formal education institutions is critical, with emphasis on the learning process rather than the product.

Reviewing learned knowledge is critical in the learning process. Testing the learning content instead of restudying, which is known as the testing effect, has been demonstrated to be an effective review strategy. However, education research recommends that instructors generate practice tests, but this burdens teachers and may also hinder teaching quality. To resolve this issue, the current study applied a modern artificial intelligence technique (BERT) to automate the generation of tests and evaluate the testing effect through e-books in a university lecture (N = 74). The last 5 minutes of each course session were utilized to review the taught content by having students either answer cloze item questions or restudy the summary of the core concepts covered in the lecture. A reading comprehension pretest was conducted before the experiment to ensure that the differences in prior knowledge were nonsignificant between groups, and a posttest was performed to examine the effectiveness of testing. In addition, we evaluated students’ reading skills and reading engagement through their ability to identify key concepts and their interaction with e-books, respectively. A positive effect was observed for students who engaged in cloze item practice before the end of each class. The results indicated that the repeated testing group exhibited significantly better reading skills and engaged more with e-books than the restudying group did. More importantly, compared with only restudying the key concepts, answering the cloze items questions significantly improved students’ reading comprehension. Our results suggest that machine-generated cloze testing may benefit learning in higher education.

Abstract Self-direction skill is considered a vital skill for twenty-first-century learners in both the learning context and physical activity context. Analysis skill for self-directed activities requires the students to analyze their own activity data for understanding their status in that activity. It is an important phase that determines whether an appropriate plan can be set or not. This research presents a framework designed to foster students’ analysis skill in self-directed activities, which aims (1) to build a technology-enabled learning system allowing students to practice analyzing data from their own daily contexts, (2) to propose an approach to model student’s analysis skill acquisition level and process, and (3) to provide automated support and feedback for analysis skill development tasks. The analysis module based on the proposed framework was implemented in the GOAL system which synchronized data from learners’ physical and reading activities. A study was conducted with 51 undergraduate students to find reliable indicators for the model to then measure students’ analysis skills. By further analyzing students’ actual usage of the GOAL system, we found the actual activity levels and their preferences regarding analysis varied for the chosen contexts (learning and physical activity). The different context preference groups were almost equal, highlighting the utility of a system that integrates data from multiple contexts. Such a system can potentially respond to students’ individual preferences to execute and acquire self-direction skill.

Research has revealed the positive effects of flipped classroom approaches on students’ learning engagement and performance compared with conventional lecture-based classrooms. However, because of a lack of out-of-class learning support, many students fail to comprehensively prepare the provided lecture materials before class. One promising solution to this problem is recommendation systems in the educational area, which have been instrumental in helping learners identify useful and relevant lecture materials that satisfy their learning needs. Thus, in this study, we propose a summarized lecture material recommendation system, which is integrated into an e-book reading system as an enhancement of the flipped classroom approach. This system helps students identify pages that contain essential knowledge that must be thoroughly studied before class. The proposed system was constructed on the basis of our previous work. In this study, a quasi-experiment was conducted in a graduate course that implemented the flipped classroom model: experimental group students learned with the proposed system, whereas the control group students had no access to the additional features. The findings of this study suggest that students who learn with the proposed recommendation system significantly outperform those who learn without the system in a flipped classroom in terms of their learning outcomes and engagement in preclass preparation.

Precision education is a new challenge in leveraging artificial intelligence, machine learning, and learning analytics to enhance teaching quality and learning performance. To facilitate precision education, text marking skills can be used to determine students’ learning process. Text marking is an essential learning skill in reading. In this study, we proposed a model that leverages the state-of-the-art text summarization technique, Bidirectional Encoder Representations from Transformers (BERT), to calculate the marking score for 130 graduate students enrolled in an accounting course. Then, we applied learning analytics to analyze the correlation between their marking scores and learning performance. We measured students’ self-regulated learning (SRL) and clustered them into four groups based on their marking scores and marking frequencies to examine whether differences in reading skills and text marking influence students’ learning performance and awareness of self-regulation. Consistent with past research, our results did not indicate a strong relationship between marking scores and learning performance. However, high-skill readers who use more marking strategies perform better in learning performance, task strategies, and time management than high-skill readers who use fewer marking strategies. Furthermore, high-skill readers who actively employ marking strategies also achieve superior scores of environment structure, and task strategies in SRL than low-skill readers who are inactive in marking. The findings of this research provide evidence supporting the importance of monitoring and training students’ text marking skill and facilitating precision education.

Computerized adaptive testing (CAT) can effectively facilitate student assessment by dynamically selecting questions on the basis of learner knowledge and item difficulty. However, most CAT models are designed for one-time evaluation rather than improving learning through formative assessment. Since students cannot remember everything, encouraging them to repeatedly evaluate their knowledge state and identify their weaknesses is critical when developing an adaptive formative assessment system in real educational contexts. This study aims to achieve this goal by proposing an adaptive formative assessment system based on CAT and the learning memory cycle to enable the repeated evaluation of students' knowledge. The CAT model measures student knowledge and item difficulty, and the learning memory cycle component of the system accounts for students’ retention of information learned from each item. The proposed system was compared with an adaptive assessment system based on CAT only and a traditional nonadaptive assessment system. A 7-week experiment was conducted among students in a university programming course. The experimental results indicated that the students who used the proposed assessment system outperformed the students who used the other two systems in terms of learning performance and engagement in practice tests and reading materials. The present study provides insights for researchers who wish to develop formative assessment systems that can adaptively generate practice tests.

Previous studies have found that the frequency and the regularity of taking the self-assessment are positively correlated with learning performance. However, as artificial intelligence is widely used for self-assessment in various educational contexts, numerous behaviors have been identified, including nonstandard behaviors that can negatively impact learning. Therefore, more analysis regarding students' self-assessment behaviors in different contexts and their influence on learning is required. In this study, we examined students’ behaviors in online self-assessment task and how it affects their learning performance. A 6-week experiment was conducted in an accounting course. Students were instructed to complete online self-assessment in the form of formative quizzes after class. We performed clustering analysis, which revealed three behavioral patterns in online self-assessment, and we compared the learning performance of students across different patterns. The results indicated that students who frequently took the online assessments after class tended to achieve a higher examination score than those who did not. However, the learning performance of students who demonstrated nonstandard behaviors did not necessarily improve, even though they actively took the assessments. Our results suggest that student behavior is a critical factor in improving learning through self-assessment. These findings provide insights for researchers in the learning analytics field as well as for practitioners who wish to adopt online self-assessment for learning.

UNESCO reported that 90% of students are affected in some way by COVID-19 pandemic. Like many countries, Japan too imposed emergency remote teaching and learning at both school and university level. In this study, we focus on a national university in Japan, and investigate how teaching and learning were facilitated during this pandemic period using an ebook platform, BookRoll, which was linked as an external tool to the university’s learning management system. Such an endeavor also reinforced the Japanese national thrust regarding explorations of e-book-based technologies and using Artificial Intelligence in education. Teachers could upload reading materials for instance their course notes and associate an audio of their lecture. While students who registered in their course accessed the learning materials, the system collected their interaction logs in a learning record store. Across the spring semesters from April - July 2020, BookRoll system collected nearly 1.5 million reading interaction logs from more than 6300 students across 243 courses in 6 domains. The analysis highlighted that during emergency remote teaching and learning BookRoll maintained a weekly average traffic above 1,900 learners creating more than 78,000 reading logs and teachers perceived it as useful for orchestrating their course.

Abstract A growing number of language learners use ubiquitous language learning applications to learn anytime and anywhere. Learners translate and learn isolated words inspired by their activities and surroundings. However, isolated words may have several meanings that change depending on the context. Since learners don’t have the opportunity to indicate the meaning they are looking for in an online learning environment, they risk learning translations that do not correspond to their intended meaning. Identifying the intended meaning of the learner is needed to provide them with an appropriate translation. However, isolated words are difficult to disambiguate due to a lack of text around them. To this end, informal ubiquitous learning environments can offer another type of context, one that is formed by the users’ past learning logs. In this work, we propose using the learners’ past vocabulary to disambiguate their intended meaning when they look up isolated words. Accordingly, we propose and evaluate three methods. The first method considers that the intended meaning of the learner is the one that is the most semantically similar to the learner’s past vocabulary. The second method builds on the first method but gives more weight to the vocabulary that the learner logged shortly before the target word. The third method addresses situations where the semantic similarities between the different meanings of the word and the past vocabulary have similar values. In those cases, the method considers that the intended meaning of the learner is the most common meaning in the target language. The three methods were evaluated using 148 logs of SCROLL, a ubiquitous informal language learning environment. The success rates of the three methods were 72.180%, 75.630%, and 83.050% respectively. This work shows that the past activity of language learners in informal ubiquitous language learning environments could be used to identify their intended meaning when learning a new word.

Abstract Early prediction systems have already been applied successfully in various educational contexts. In this study, we investigated developing an early prediction system in the context of eBook-based teaching-learning and used students’ eBook reading data to develop an early warning system for students at-risk of academic failure -students whose academic performance is low. To determine the best performing model and optimum time for possible interventions we created prediction models by using 13 prediction algorithms with the data from different weeks of the course. We also tested effects of data transformation on prediction models. 10-fold cross-validation was used for all prediction models. Accuracy and Kappa metrics were used to compare the performance of the models. Our results revealed that in a sixteen-week long course all models reached their highest performance with the data from the 15th week. On the other hand, starting from the 3rd week, the models classified low and high performing students with an accuracy of over 79%. In terms of algorithms, Random Forest (RF) outperformed other algorithms when raw data were used, however, with the transformed data J48 algorithm performed better. When categorical data were used, Naive Bayes (NB) outperformed other algorithms. Results also indicated that models with transformed data performed lower than the models created using categorical data. However, models with categorical data showed similar performance with models with raw data. The implications of the results presented in this research were also discussed with respect to the field of Learning Analytics.

Abstract It is a common practice to issue a summary of a learner’s learning achievements in form of a transcript or certificate. However, detailed information on the depth of learning and how learning or teachings were conducted is not present in the transcript of scores. This work presents the first practical implementation of a new platform for keeping track of learning achievements beyond transcripts and certificates. This is achieved by maintaining digital hashes of learning activities and managing access rights through the use of smart contracts on the blockchain. The blockchain of learning logs (BOLL) is a platform that enable learners to move their learning records from one institution to another in a secure and verifiable format. This primarily solves the cold-start problem faced by learning data analytic platforms when trying to offer personalized experience to new learners. BOLL enables existing learning data analytic platforms to access the learning logs from other institutions with the permission of the learners and/or institution who originally have ownership of the logs. The main contribution of this paper is to investigate how learning records could be connected across institutions using BOLL. We present an overview of how the implementation has been carried out, discuss resource requirements, and compare the advantages BOLL has over other similar tools.

Abstract The aim of this research is to measure self-regulated behavior and identify significant behavioral indicators in computer-assisted language learning courses. The behavioral measures were based on log data from 2454 freshman university students from Art and Science departments for 1 year. These measures reflected the degree of self-regulation, including anti-procrastination, irregularity of study interval, and pacing. Clustering analysis was conducted to identify typical patterns of learning pace, and hierarchical regression analysis was performed to examine significant behavioral indicators in the online course. The results of learning pace clustering analysis revealed that the final course point average in different clusters increased with the number of completed quizzes, and students who had procrastination behavior were more likely to achieve lower final course points. Furthermore, the number of completed quizzes and study interval irregularity were strong predictors of course performance in the regression model. It clearly indicated the importance of self-regulation skill, in particular completion of assigned tasks and regular learning.

Abstract E-book reader supports users to create digital learning footprints in many forms like highlighting sentences or taking memos. Nowadays, it also allows an instructor to update their e-books in the e-book reader. However, e-book users often face problems when trying to find learning footprints they made in a new version e-book. Thus, users’ reading experience continuity across e-book revisions is hard to be maintained and seems to become a shortcoming within the e-book system. In this paper, in order to maintain users’ reading experience continuity, we deal with the transfer of learning footprints such as a marker, memo, and bookmark across e-book revisions on an e-book reader in a coursework scenario. We first give introduction and related works to demonstrate how researchers dedicated on the problem mentioned in this paper and page similarity comparison. Then, we compare three page similarity comparison methods using similarity computing models to compute page pairwise similarity in image level, text level, and image & text level. In the analysis, for each level, we analyze the performance of transferring learning footprint across e-book revisions and also the optimal threshold for similar page determination. After that, we give the analysis results to show the performances of three methods in image level, text level, and image & text level, and then, the error analysis is presented to specify the error types that occur in the results. We then propose page image & text similarity comparison as the optimal method to automatically transfer learning footprints across e-book revisions based on the analysis results and error analysis among three compared methods. Finally, the discussion and conclusions are shown in the end of this paper.

In recent years, learning analytics has become a hot topic with many institutes deploying learning management systems and learning analytics tools. In this paper, we introduce learning analytics platforms that have been established in two top national Japanese universities. These initiatives are part of a broader research project into creating wide-reaching learning analytics frameworks. The aim of the project is to support education and learning through research into educational big data accumulated on these platforms. We also discuss the future direction of our research into learning analytics platforms. This includes introducing a model in which learning analytics tools and the results of research can be shared between different education institutes.

Abstract Successful language learning requires an understanding of the target culture in order to make valuable usage of the learned language. To understand a foreign culture, language students need the knowledge of its related products, as well as the skill of comparing them to those of their own culture. One way for students to understand foreign products is by making Culturally Situated Associations (CSA), i.e., relating the products they encountered to products from their own culture. In order to provide students with CSA that they can understand, we must gather information about their culture, provide them with the CSA, and make sure they understand it. In this case, a Culturally Situated Dialogue (CSD) must take place. To carry the dialogue, dialogue systems must follow a dialogue strategy. However, previous work showed that handcrafted dialogue strategies were shown to be ineffective in comparison with machine-learned dialogue strategies. In this research, we proposed a method to learn CSD strategies to support foreign students, using a reinforcement learning algorithm. Since no previous system providing CSA was implemented, the method allowed the creation of CSD strategies when no initial data or prototype exists. The method was applied to generate three different agents: the novice agent was based on an eight states feature-space, the intermediate agent was based on a 144 states feature-space, and the advanced agent was based on a 288 states feature-space. Each of these agents learned a different dialogue strategy. We conducted a Wizard of Oz experiment during which, the agents’ role was to support the wizard in their dialogue with students by providing them with the appropriate action to take at each step. The resulting dialogue strategies were evaluated based on the quality of the strategy. The results suggest the use of the novice agent at the first stages of prototyping the dialogue system. The intermediate agent and the advanced agent could be used at later stages of the system’s implementation.

BackgroundLittle is known about the utilization of cardiac diagnostic testing in Brazil and how such testing is related with local rates of acute coronary syndrome (ACS)-related mortality.Methods and resultsUsing data from DATASUS, the public national healthcare database, absolute counts of diagnostic tests performed were calculated for each of the 5570 municipalities and mapped. Spatial error regression and geographic weighted regression models were used to describe the geographic variation in the association between ACS mortality, income, and access to diagnostic testing. From 2008 to 2014, a total of 4,653,884 cardiac diagnostic procedures were performed in Brazil, at a total cost of $271 million USD. The overall ACS mortality rate during this time period was 133.8 deaths per 100,000 inhabitants aged 20 to 79. The most commonly utilized test was the stress ECG (3,015,993), followed by catheterization (862,627), scintigraphy (669,969) and stress echocardiography (105,295). The majority of these procedures were conducted in large urban centers in more economically developed regions of the country. Increased access to testing and increased income were not uniformly associated with decreased ACS mortality, and tremendous geographic heterogeneity was observed in the relationship between these variables.ConclusionsThe majority of testing for ACS in Brazil is conducted at referral centers in developed urban settings. Stress ECG is the dominant testing modality in use. Increased access to diagnostic testing was not consistently associated with decreased ACS mortality across the country.

A large number of refugees need to learn a new language in order to transition to, or settle in, a new country. Most refugees have access to smartphones with which they can access mobile language learning tools. However, refugees in different stages of their journeys have different language learning needs. Identifying the unique needs of refugees is essential to design language learning tools that support their language learning activities. Through a two-part participatory design study, we identify and compare the tacit and latent language learning needs of two different groups of Syrian refugees: a group in transition in Lebanon, and a group settling in Germany. The Syrian refugees we worked with in Lebanon are studying English to transition to another country with better living conditions. On the other hand, the Syrian refugees we worked with in Germany are trying to learn German to better integrate into German society. Our results show overlapping needs and specific needs for the two groups of refugees. We present our findings and discuss the opportunities and challenges for designing language learning tools to support the language learning activities of refugees.

Self-directed learning (SDL) requires students to take initiative to learn and control their own learning process. Literature highlights the importance of SDL for lifelong learning. Yet, little understanding is known regarding how to support SDL at the school level, specifically for out-of-class learning context. To fill up this gap, this research developed a learning dashboard and integrated SDL process management in GOAL system. It was implemented to provide support for out-of-class online self-directed extensive reading (SDER) at a high school level. A two-group study conducted during a three-week spring vacation found the experimental group (N = 35, with SDL support) achieved significantly more reading outcomes than the control group (N = 12, without SDL support). Detailed GOAL interaction behavior analysis of the experimental group showed that viewing learning dashboard was significantly correlated with reading outcomes as well as interactions related to SDL process management. These findings highlights positive effect of SDL support in GOAL on students' out-of-class SDER outcomes as well as their SDL behaviors. The study provided implications for research related to extensive reading and SDL support for out-of-class learning.

Explainable recommendation, which provides an explanation about why a quiz is recommended, helps to improve transparency, persuasiveness, and trustworthiness. However, little research examined the effectiveness of the explainable recommender, especially on academic performance. To survey its effectiveness, the authors evaluate the math academic performance among middle school students (n=115) by giving pre- and post-test questions based evaluation techniques. During the pre- and post-test periods, students were encouraged to use the Bayesian Knowledge Tracing model based explainable recommendation system. To evaluate how well the students were able to do what they could not do, the authors defined growth rate and found recommended quiz clicked counts had a positive effect on the total number of solved quizzes (R=0.343, P=0.005) and growth rate (R=0.297, P=0.017) despite no correlation between the total number of solved quizzes and growth rate. The results suggest that the use of an explainable recommendation system that learns efficiently will enable students to do what they could not do before.

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.

The use of blockchain in education has become one of the trending topics in education technology research. However, only a handful of education blockchain solutions have provided a measure of the impact on students' learning outcomes, teaching, or administrative processes. This work reviews how academic data stored on the blockchain is being visualized across various education blockchain solutions. We argue that education's uniqueness requires a different visualization approach that supports students' learning activities, advances teaching methods, and facilitates administrative procedures. We identify a consistent trend where most of the proposed education blockchain solutions focus on credentials collection and do not provide a way to make sense of the blockchain's data. Thus, we conducted a needs analysis by interviewing four teachers to understand essential features when accessing distributed academic data, report these results and use them to inform the features of our proposed visualizations. Our unique contributions include: presenting typical use cases of distributed learning records from multiple education institutes and demonstrating how past learning records of students stored on the blockchain can be visualized to support current learning. We also propose a method of visualization to increase the data awareness of information owners through the blockchain.