Automated Essay Scoring (AES) research efforts primarily focus on feature engineering and the building of machine learning models to attain higher consensus with human graders. In academic grading such as essay scoring, the scores will naturally result in a normal distribution, more commonly referred to as the bell curve. However, the datasets used do not always have such distribution and are often overlooked in most machine learning environments. This paper proposes a Gaussian Multi-Class Synthetic Minority Over-sampling Technique (GMC-SMOTE) for imbalanced datasets. The proposed GMC-SMOTE generates new synthetic data to complement the existing datasets to produce scores that are in a normal distribution. Using several labeled essay sets, some of which already have a substantial agreement between the machine learning model and human graders, learning from normal distribution datasets yields significant improvements. Improvements of 0.038 QWK score (5.8%) over the imbalanced dataset were observed. The experimental result has also shown that naturally occurring distribution in the automated essay scoring domain contributes to the most appropriate training dataset for machine learning purposes. © 2022 Copyright is held by the author(s).

To improve student learning outcomes within online learning platforms, struggling students are often provided with on-demand supplemental instructional content. Recently, services like Yup (yup.com) and UPcheive (upchieve.org) have begun to offer on-demand live tutoring sessions with qualified educators, but the availability of tutors and the cost associated with hiring them prevents many students from having access to live support. To help struggling students and offset the inequities intrinsic to high-cost services, we are attempting to develop a process that uses large language representation models to algorithmically identify relevant support messages from these chat logs, and distribute them to all students struggling with the same content. In an empirical evaluation of our methodology we were able to identify messages from tutors to students struggling with middle school mathematics problems that qualified as explanations of the content. However, when we distributed these explanations to students outside of the tutoring sessions, they had an overall negative effect on the students’ learning. Moving forward, we want to be able to identify messages that will promote equity and have a positive impact on students. © 2022 Copyright is held by the author(s).

As computer-based learning platforms have become ubiquitous, there is a growing need to better support teachers. Particularly in mathematics, teachers often rely on open-ended questions to assess students’ understanding. While prior works focusing on the development of automated open-ended work assessments have demonstrated their potential, many of those methods require large amounts of student data to make reliable estimates. We explore whether a problem specific automated scoring model could benefit from auxiliary data collected from similar problems to address this “cold start” problem. We examine factors such as sample size and the magnitude of similarity of utilized problem data. We find the use of data from similar problems not only provides benefits to improve predictive performance by increasing sample size, but also leads to greater overall model performance than using data solely from the original problem when sample size is held constant. © 2022 Copyright is held by the author(s).

Advancements in online learning platforms have revolutionized education in multiple different ways, transforming the learning experiences and instructional practices. The development of natural language processing and machine learning methods have helped understand and process student languages, comprehend their learning state, and build automated supports for teachers. With this, there has been a growing body of research in developing automated methods to assess students’ work both in mathematical and non-mathematical domains. These automated methods address questions of two categories; closed-ended (with limited correct answers) and open-ended (are often subjective and have multiple correct answers), where open-ended questions are mostly used by teachers to learn about their student’s understanding of a particular concept. Manually assessing and providing feedback to these open-ended questions is often arduous and time-consuming for teachers. For this reason, there have been several works to understand student responses to these open-ended questions to automate the assessment and provide constructive feedback to students. In this research, we seek to improve such a prior method for assessment and feedback suggestions for student open-ended works in mathematics. For this, we present an error analysis of the prior method”SBERT-Canberra” for auto-scoring, explore various factors that contribute to the error of the method, and propose solutions to improve upon the method by addressing these error factors. We further intend to expand this approach by improving feedback suggestions for teachers to give to their students’ open-ended work. © 2022 Copyright is held by the author(s).

As online learning platforms become more ubiquitous throughout various curricula, there is a growing need to evaluate the effectiveness of these platforms and the different methods used to structure online education and tutoring. Towards this endeavor, some platforms have performed randomized controlled experiments to compare different user experiences, curriculum structures, and tutoring strategies in order to ensure the effectiveness of their platform and personalize the education of the students using it. These experiments are typically analyzed on an individual basis in order to reveal insights on a specific aspect of students’ online educational experience. In this work, the data from 50,752 instances of 30,408 students participating in 50 different experiments conducted at scale within the online learning platform ASSISTments were aggregated and analyzed for consistent trends across experiments. By combining common experimental conditions and normalizing the dependent measures between experiments, this work has identified multiple statistically significant insights on the impact of various skill mastery requirements, strategies for personalization, and methods for tutoring in an online setting. This work can help direct further experimentation and inform the design and improvement of new and existing online learning platforms. The anonymized data compiled for this work are hosted by the Open Science Foundation and can be found at https://osf.io/59shv/. © 2022 Copyright is held by the author(s).

Studies have proven that providing on-demand assistance, additional instruction on a problem when a student requests it, improves student learning in online learning environments. Additionally, crowdsourced, on-demand assistance generated from educators in the field is also effective. However, when provided on-demand assistance in these studies, students received assistance using problem-based randomization, where each condition represents a different assistance, for every problem encountered. As such, claims about a given educator's effectiveness are provided on a per-assistance basis and not easily generalizable across all students and problems. This work aims to provide stronger claims on which educators are the most effective at generating on-demand assistance. Students will receive on-demand assistance using educator-based randomization, where each condition represents a different educator who has generated a piece of assistance, allowing students to be kept in the same condition over longer periods of time. Furthermore, this work also attempts to find additional benefits to providing students assistance generated by the same educator compared to a random assistance available for the given problem. All data and analysis being conducted can be found on the Open Science Foundation website (https://osf.io/zcbjx/).

The development and application of deep learning methodologies has grown within educational contexts in recent years. Perhaps attributable, in part, to the large amount of data that is made available through the adoption of computer-based learning systems in classrooms and larger-scale MOOC platforms, many educational researchers are leveraging a wide range of emerging deep learning approaches to study learning and student behavior in various capacities. Variations of recurrent neural networks, for example, have been used to not only predict learning outcomes but also to study sequential and temporal trends in student data; it is commonly believed that they are able to learn high-dimensional representations of learning and behavioral constructs over time, such as the evolution of a students' knowledge state while working through assigned content. Recent works, however, have started to dispute this belief, instead finding that it may be the model's complexity that leads to improved performance in many prediction tasks and that these methods may not inherently learn these temporal representations through model training. In this work, we explore these claims further in the context of detectors of student affect as well as expanding on existing work that explored benchmarks in knowledge tracing. Specifically, we observe how well trained models perform compared to deep learning networks where training is applied only to the output layer. While the highest results of prior works utilizing trained recurrent models are found to be superior, the application of our untrained-versions perform comparably well, outperforming even previous non-deep learning approaches.

Many teachers have come to rely on the affordances that computer-based learning platforms offer in regard to aiding in student assessment, supplementing instruction, and providing immediate feedback and help to students as they work through assigned content. Similarly, researchers commonly utilize the large datasets of clickstream logs describing students' interactions with the platform to study learning. For the teachers that use this information to monitor student progress, as well as for researchers, this data provides limited insights into the learning process; this is particularly the case as it pertains to observing and understanding the effort that students are applying to their work. From the perspective of teachers, it is important for them to know which students are attending to and using computer-provided aid and which are taking advantage of the system to complete work without effectively learning the material. In this paper, we conduct a series of analyses based on response time decomposition (RTD) to explore student help-seeking behavior in the context of on-demand hints within a computer-based learning platform with particular focus on examining which students appear to be exhibiting effort to learn while engaging with the system. Our findings are then leveraged to examine how our measure of student effort correlates with later student performance measures.

Despite the abundance of data generated from students' activities in virtual learning environments, the use of supervised machine learning in learning analytics is limited by the availability of labeled data, which can be difficult to collect for complex educational constructs. In a previous study, a subfield of machine learning called Active Learning (AL) was explored to improve the data labeling efficiency. AL trains a model and uses it, in parallel, to choose the next data sample to get labeled from a human expert. Due to the complexity of educational constructs and data, AL has suffered from the cold-start problem where the model does not have access to sufficient data yet to choose the best next sample to learn from. In this paper, we explore the use of past data to warm start the AL training process. We also critically examine the implications of differing contexts (urbanicity) in which the past data was collected. To this end, we use authentic affect labels collected through human observations in middle school mathematics classrooms to simulate the development of AL-based detectors of engaged concentration. We experiment with two AL methods (uncertainty sampling, L-MMSE) and random sampling for data selection. Our results suggest that using past data to warm start AL training could be effective for some methods based on the target population's urbanicity. We provide recommendations on the data selection method and the quantity of past data to use when warm starting AL training in the urban and suburban schools.

As more educators integrate their curricula with online learning, it is easier to crowdsource content from them. Crowdsourced tutoring has been proven to reliably increase students' next problem correctness. In this work, we confirmed the findings of a previous study in this area, with stronger confidence margins than previously, and revealed that only a portion of crowdsourced content creators had a reliable benefit to students. Furthermore, this work provides a method to rank content creators relative to each other, which was used to determine which content creators were most effective overall, and which content creators were most effective for specific groups of students. When exploring data from TeacherASSIST, a feature within the ASSISTments learning platform that crowdsources tutoring from teachers, we found that while overall this program provides a benefit to students, some teachers created more effective content than others. Despite this finding, we did not find evidence that the effectiveness of content reliably varied by student knowledge-level, suggesting that the content is unlikely suitable for personalizing instruction based on student knowledge alone. These findings are promising for the future of crowdsourced tutoring as they help provide a foundation for assessing the quality of crowdsourced content and investigating content for opportunities to personalize students' education. [This paper was published in: " L@S '21, June 22-25, 2021, Virtual Event, Germany," ACM, 2021.]