Effective file management is central to coordination in collaborative work, as shared files serve as the primary medium through which collaborators exchange contributions. Building on existing PIM and CSCW literature on file management breakdowns, we recontextualize such breakdowns within specific dynamics of collaboration. In Study 1, we conducted a need-finding interview(N=33) and identified four recurring breakdowns in collaborative file management: ambiguous file placement and ownership, inefficient version management, uninterpretable metadata, and missing status cues. Building on these findings, Study 2 used a design probe evaluation(N=12) to examine potential benefits and concerns associated with supporting collaborative file management. Participants reported benefits such as clearer ownership, stronger reference convergence, improved metadata interpretability, and heightened progress visibility, while expressing concerns related to surveillance, exploration containment, overdisclosure, and social pressure. Taken together, the studies reframe well-known file management issues as a dichotomy between perceived benefits and concerns, thereby outlining design directions for file-level alignment.

This quasi-experimental study integrates a large language model (LLM) with expert qualitative analysis to examine how instructional design variations in computer-supported collaborative learning (CSCL) shape collaboration in data-centric programming. We collected 73 team transcripts from two contrasting CSCL designs deployed across five course offerings: a closed-ended variant with prescribed solution paths and auto-graded milestones, and an open-ended variant supporting exploratory tasks with multiple valid paths. LLM annotation revealed statistically significant differences in knowledge co-construction patterns: the open-ended design yielded a higher proportion of utterances focused on developing a shared understanding of problems and solutions. Guided by these quantitative results, human experts conducted qualitative coding that confirmed and enriched these findings, showing how open-ended tasks fostered elaborative solution negotiation while closed-ended structures promoted non-elaborative exchanges. Our contributions are: (1) instructional insights for data science education, demonstrating how open-ended CSCL designs better support collaborative sense-making essential for real-world data science projects; and (2) a documented workflow for human-LLM co-evaluation, providing the methodological detail necessary for others to replicate our process and apply it to future studies.

We introduce game-player agency, defined as a student’s desire and ability to determine actions related to game elements in game-based learning environments. Game-player agency differs from student agency as it is more closely associated with game elements than with the learning process. Our study investigates (1)the relationship between game-player agency and student agency, (2)the relationships between both forms of agency and learning gain, (3)the relationships between both forms of agency and engagement, and (4)the indirect effects of engagement on the relationship between agency and learning gain. Log data from 2,261 students in an English learning game, LingoCity, was analyzed across three different age groups; earlier elementary grades (aged 7-9), later elementary grades (aged 10-12), and middle school grades (aged 13-15). Our results showed that (1)an inverted U-shaped relationship exists between game-player and student agency in all groups (p-values <0.05); (2)for middle school grades, learning gain had a positive relationship with student agency (p=0.030) and a negative relationship with game-player agency ($\mathbf{p}{\lt}0.001$), while no significant direct effects were observed for elementary grades; (3)both agencies had a positive relationship with engagement, except for student agency and total engagement frequency—how often students interacted with the learning game regardless of activity type—in earlier elementary grades; and (4)learning engagement frequency—how often students participated in learning-related activities—significantly mediated both agencies and learning gain in all groups. The results underscore the importance of game-player agency, showing its positive link to student agency and strong ties to engagement and learning gain.

Audio-based exergames are beneficial in that they allow users to exercise in an eyes-free and hands-free environment. In this study, we explored two audio-based exergame elements, narrative features and thematic music, that can impact exercise amount (step count and duration) and exercise enjoyment. We conducted a two-week long between-subjects study with 43 young adults and 43 middle-aged adults using SPORTIFY, an audio-based exergame. Our experimental results showed that (1) Using narrative features had a significant main effect on exercise amount and exercise enjoyment both for young adults and middle-aged adults. (2) Using thematic music had no significant main impact on exercise amount and exercise enjoyment both for young adults and middle-aged adults. (3) A significant interaction effect for exercise amount was observed in middle-aged adults, whereas a significant interaction effect for exercise enjoyment was observed in young adults.

Herein, we present two studies on how students’ Psychological State of Decision difficulty (PSD) relates to two aspects of learning, i.e., guessing behavior and learning achievement. To measure PSD, we extracted geometric features from trajectories of drag-and-drop touch interactions collected while students aged 7–10 played a math game on a tablet device. In the first study, we explored whether eight geometric features extracted from 97,303 trial trajectories could be grouped to understand students’ PSD. In the second study, we examined whether the two aspects of learning could be predicted using the data collected from 187 students with geometric features indicating their PSD. This work provides empirical evidence that geometric features can be grouped into two types of PSD in the context of learning, including conflict and uncertainty. Moreover, our results demonstrate that data on students’ PSD collected from drag-and-drop trajectories can be used to predict learning.

In this paper, we propose a neural model EPT-X (Expression-Pointer Transformer with Explanations), which utilizes natural language explanations to solve an algebraic word problem.To enhance the explainability of the encoding process of a neural model, EPT-X adopts the concepts of plausibility and faithfulness which are drawn from math word problem solving strategies by humans.A plausible explanation is one that includes contextual information for the numbers and variables that appear in a given math word problem.A faithful explanation is one that accurately represents the reasoning process behind the model's solution equation.The EPT-X model yields an average baseline performance of 69.59% on our PEN dataset and produces explanations with quality that is comparable to human output.The contribution of this work is two-fold.(1) EPT-X model: An explainable neural model that sets a baseline for algebraic word problem solving task, in terms of model's correctness, plausibility, and faithfulness.(2) New dataset: We release a novel dataset PEN (Problems with Explanations for Numbers), which expands the existing datasets by attaching explanations to each number/variable.

In this paper, we extend existing research on using geometric features extracted from trajectory data to understand student behavioral patterns and learning outcome. We analyzed 910,661 trajectories data from 390 students who played the KitKit School, a mathematics educational game. As a result of factor analysis, three behavioral patterns are computed: conflict, wavering, and locomotion. A conflict pattern is the degree of curvature of trajectory and implies the degree of decision conflict. A wavering pattern is the number of times when the directions of a trajectory change and refers to the level of confusion a user may feel between choices. Locomotion pattern is the trajectory length of a user's movement while making a choice. The correlation analysis results show that conflict (r=-0.145, p=0.004) and wavering (r=-0.100, p=0.049) negatively correlated with the learning outcome. There is no significant correlation between locomotion and learning outcome (r=0.076, p=0.133). The contributions of this paper are (1) Identification of three types of student behavioral patterns using geometric features of trajectories: conflict, wavering, and locomotion (2) Findings on a negative relationship between learning outcome and the two types of behavioral patterns, conflict and wavering.

Securing sufficient data to enable automatic sign language translation modeling is challenging. The data insufficiency issue exists in both video and text modalities; however, fewer studies have been performed on text data augmentation compared to video data. In this study, we present three methods of augmenting sign language text modality data, comprising 3,052 Gloss-level Korean Sign Language (GKSL) and Word-level Korean Language (WKL) sentence pairs. Using each of the three methods, the following number of sentence pairs were created: blank replacement 10,654, sentence paraphrasing 1,494, and synonym replacement 899. Translation experiment results using the augmented data showed that when translating from GKSL to WKL and from WKL to GKSL, Bi-Lingual Evaluation Understudy (BLEU) scores improved by 0.204 and 0.170 respectively, compared to when only the original data was used. The three contributions of this study are as follows. First, we demonstrated that three different augmentation techniques used in existing Natural Language Processing (NLP) can be applied to sign language. Second, we propose an automatic data augmentation method which generates quality data by utilizing the Korean sign language gloss dictionary. Lastly, we publish the Gloss-level Korean Sign Language 13k dataset (GKSL13k), which has verified data quality through expert reviews.

Backchannel (BC), a short reaction signal of a listener to a speaker's utterances, helps to improve the quality of the conversation.Several studies have been conducted to predict BC in conversation; however, the utilization of advanced natural language processing techniques using lexical information presented in the utterances of a speaker has been less considered.To address this limitation, we present a BC prediction model called BPM_MT (Backchannel prediction model with multitask learning), which utilizes KoBERT, a pre-trained language model.The BPM_MT simultaneously carries out two tasks at learning: 1) BC category prediction using acoustic and lexical features, and 2) sentiment score prediction based on sentiment cues.BPM_MT exhibited 14.24% performance improvement compared to the existing baseline in the four BC categories: continuer, understanding, empathic response, and No BC.In particular, for empathic response category, a performance improvement of 17.14% was achieved.

We examined the effects of the reading medium (print vs. digital) on readers’ visual patterns, reading performance, and reading attitudes. Two within-subject experiments were conducted with 74 readers, who read articles using three reading media: print, computer, and tablet. The experimental results showed that in terms of visual patterns, readers exhibited a shorter fixation duration and a higher fixation count during print reading than during screen reading; reading performance, as measured on the basis of reading comprehension and reading time, was equivalent across all three media; however, in terms of reading attitude, readers reported higher levels of perceived understanding, perceived confidence, and perceived immersion and lower levels of perceived fatigue for reading printed text than reading from a device screen. Therefore, the performance gap between print and screen reading is narrowing. However, printed text may still be the preferred mode of reading, as demonstrated by the readers’ preferences.

Math word problem solving is an emerging research topic in Natural Language Processing.Recently, to address the math word problem solving task, researchers have applied the encoderdecoder architecture, which is mainly used in machine translation tasks.The state-of-the-art neural models use hand-crafted features and are based on generation methods.In this paper, we propose the GEO (Generation of Equations by utilizing Operators) model that does not use handcrafted features and addresses two issues that are present in existing neural models: 1. missing domain-specific knowledge features and 2. losing encoder-level knowledge.To address missing domain-specific feature issue, we designed two auxiliary tasks: operation group difference prediction and implicit pair prediction.To address losing encoder-level knowledge issue, we added an Operation Feature Feed Forward (OP3F) layer.Experimental results showed that the GEO model outperformed existing state-of-the-art models on two datasets, 85.1% in MAWPS, and 62.5% in DRAW-1K, and reached comparable performance of 82.1% in ALG514 dataset.

Solving algebraic word problems has recently emerged as an important natural language processing task.To solve algebraic word problems, recent studies suggested neural models that generate solution equations by using 'Op (operator/operand)' tokens as a unit of input/output.However, such a neural model suffered two issues: expression fragmentation and operand-context separation.To address each of these two issues, we propose a pure neural model, Expression-Pointer Transformer (EPT), which uses (1) 'Expression' token and (2) operand-context pointers when generating solution equations.The performance of the EPT model is tested on three datasets: ALG514, DRAW-1K, and MAWPS.Compared to the state-of-the-art (SoTA) models, the EPT model achieved a comparable performance accuracy in each of the three datasets; 81.3% on ALG514, 59.5% on DRAW-1K, and 84.5% on MAWPS.The contribution of this paper is two-fold; (1) We propose a pure neural model, EPT, which can address the expression fragmentation and the operandcontext separation.(2) The fully automatic EPT model, which does not use hand-crafted features, yields comparable performance to existing models using hand-crafted features, and achieves better performance than existing pure neural models by at most 40%.

We investigated the similarities and differences among four addiction groups in Korean adolescents: Non-Addiction (NONE), Smartphone Addiction (SA), Internet Addiction (IA), and Internet-Smartphone Addiction (BOTH). For the dependent variables, we examined 12 addiction-risk factors related to psychology, family, and school environment that can influence the adolescents’ normative developmental path. To collect data, we conducted an addiction-risk factor comparison survey with 768 Korean adolescents in their first year of junior high school. Depending on the addiction groups, a multivariate analysis of variance or Tukey HSD post-hoc test was used to analyze statistical differences among the 12 addiction-risk factors. Our analysis yielded two key findings on how Internet addiction and smartphone addiction differ in terms of addiction-risk factors: (a) there were more similarities between the SA and NONE groups than between the IA and NONE groups, and (b) there were more similarities between the IA and BOTH groups than between the SA and BOTH groups.

Abstract With an increase in the number of Web documents, the number of proposed methods for knowledge discovery on Web documents have been increased as well. The documents do not always provide keywords or categories, so unsupervised approaches are desirable, and topic modeling is such an approach for knowledge discovery without using labels. Further, Web documents usually have time information such as publish years, so knowledge patterns over time can be captured by incorporating the time information. The temporal patterns of knowledge can be used to develop useful services such as a graph of research trends, finding similar authors (potential co-authors) to a particular author, or finding top researchers about a specific research domain. In this paper, we propose a new topic model, Author Topic-Flow (ATF) model, whose objective is to capture temporal patterns of research interests of authors over time, where each topic is associated with a research domain. The state-of-the-art model, namely Temporal Author Topic model, has the same objective as ours, where it computes the temporal patterns of authors by combining the patterns of topics. We believe that such ‘indirect’ temporal patterns will be poor than the ‘direct’ temporal patterns of our proposed model. The ATF model allows each author to have a separated variable which models the temporal patterns, so we denote it as ‘direct’ topic flow. The design of the ATF model is based on the hypothesis that ‘direct’ topic flows will be better than the ‘indirect’ topic flows. We prove the hypothesis is true by a structural comparison between the two models and show the effectiveness of the ATF model by empirical results.

This study aims to explore the feasibility of a text-based virtual agent as a new survey method to overcome the web survey's common response quality problems, which are caused by respondents' inattention. To this end, we conducted a 2 (platform: web vs. chatbot) × 2 (conversational style: formal vs. casual) experiment. We used satisficing theory to compare the responses' data quality. We found that the participants in the chatbot survey, as compared to those in the web survey, were more likely to produce differentiated responses and were less likely to satisfice; the chatbot survey thus resulted in higher-quality data. Moreover, when a casual conversational style is used, the participants were less likely to satisfice-although such effects were only found in the chatbot condition. These results imply that conversational interactivity occurs when a chat interface is accompanied by messages with effective tone. Based on an analysis of the qualitative responses, we also showed that a chatbot could perform part of a human interviewer's role by applying effective communication strategies.

There has been a growing concern over the huge increase in use of smart media by young children. This study explores the possibility of using augmented-reality(AR) for regulat-ing preschoolers' media usage behavior. With MABLE (mobile application for behavioral learning and education), parents can provide AR-assisted feedback by changing facial expressions and sound effects. When overlaying a smart media, which has MABLE running, in front of a QR marker on a puppet, a facial expression is displayed on top of the puppet's face. A two-week long experiment with 36 parent-child pairs showed that compared to using just the puppet, using MABLE showed higher amount of engage-ment among preschoolers. For the effectiveness of parental mediation in terms of self-control, our data showed mixed results. MABLE had positive effects in that the amount of rule-compliance increased and problematic behaviors de-creased, whereas the level of behavioral dependency on smart media was not influenced.

Acquiring simple arithmetic skills at the preschool level requires repetitive practices. One method for encouraging students to spend longer time practicing is by presenting the skills in an engaging game. As student retention on the game increases, the student will be more likely to acquire the practiced skill since she will have spent more time practicing. In this paper, we examine the relationship between internal locus of control and retention in game-based learning applications for young children using Todo Math, a mobile-based math learning application for children from Pre-K to 2nd grade. We examine 345,783 users' log data to show that when children prefer "free" mode, which has high internal locus of control, their retention on Todo Math is higher than children who prefer "daily" mode, which has high external locus of control. We present three analyses that support our findings using survival analysis, post-hoc analysis, and t-test.

In this paper, we examined how the presentation of a remote participant (in our context the remote teacher) in a mobile remote presence (MRP) system affects social interaction, such as closeness and engagement. Using ROBOSEM, a MRP robot, we explored the effect of the presentation of the remote teacher's head size shown on ROBOSEM's screen at three different levels: small, medium, and large. We hypothesized that a medium sized head of the remote teacher shown on the MRP system would be better than a small or large sized head in terms of closeness, engagement, and learning. Our preliminary study results suggest that the size of a remote teacher's head may have an impact on “students' perception of the remote teacher's closeness” and on “students' engagement”. However, we did not observe any difference in terms of “learning”.

Photo sharing activities on social networking sites concern not only the person sharing the information (owner) and the person receiving the information (viewer) but also the person who is in the photo (subject). In our exploratory lab study, we asked 29 participants about their comfort level in allowing a photo owner to share a picture containing both the participant (subject) and the owner. Our results show that the photo subject feels more comfortable in sharing a photo when i) the "closeness between the subject and the owner (SO closeness)" is higher, and ii) the "closeness between the subject and the viewer (SV closeness)" is higher. In addition, we observed that both SV and SO closeness are important in determining the subject's picture sharing preference level.

The negative aspects of smartphone overuse on young adults, such as sleep deprivation and attention deficits, are being increasingly recognized recently. This emerging issue motivated us to analyze the usage patterns related to smartphone overuse. We investigate smartphone usage for 95 college students using surveys, logged data, and interviews. We first divide the participants into risk and non-risk groups based on self-reported rating scale for smartphone overuse. We then analyze the usage data to identify between-group usage differences, which ranged from the overall usage patterns to app-specific usage patterns. Compared with the non-risk group, our results show that the risk group has longer usage time per day and different diurnal usage patterns. Also, the risk group users are more susceptible to push notifications, and tend to consume more online content. We characterize the overall relationship between usage features and smartphone overuse using analytic modeling and provide detailed illustrations of problematic usage behaviors based on interview data.

Project course instructors routinely perform their formal assessments based on impressions formed from their mostly indirect experience with the groups they oversee. Nevertheless, even with their limited vantage point, instructors trust their ability to make assessments and regulate group work. In this paper we present a 5 dimensional assessment framework based on data from an interview study in which we investigate the assessment goals that project course instructors have. We use this framework to identify specifically where instructors' assessments about students diverge most from that of direct observers of group work. We then demonstrate that indicators extracted automatically from recorded speech from group meetings frequently correlate better with objective observer rating of students than that of the instructor.

While using how-to documents for guidance in performing computer-based tasks, users often run into problems due to inaccurate, out-of-date and incomplete documentation. These problems are often due to current documentation practices, which fail to keep how-to documents current, accurate, and complete. We believe that automated support for incremental update of how-to-documents, through the use of programming by demonstration and guided walkthrough techniques, is more effective than existing practice and produces documents that cause fewer problems for their users. In this paper, we present a study that evaluates this belief by comparing DocWizards, a tool utilizing these techniques, with a standard word processor. We show that more effective and efficient documentation can be generated by multiple authors using DocWizards in an incremental process, with effort comparable to that incurred using a traditional tool.

This paper describes results from a series of experimental studies to explore issues related to structuring productive group dynamics for collaborative learning using an adaptive support mechanism. The first study provides evidence in favor of the feasibility of the endeavor by demonstrating with a tightly controlled study that even without adaptive support, problem solving in pairs is significantly more effective for learning than problem solving alone. The results from a second study offer guidelines for strategic matching of students with learning partners. Furthermore, the results reveal specific areas for needed support. Based on the results from the second study, we present the design of an adaptive support mechanism, which we evaluate in a third study. The results from the third study provide evidence that certain aspects of our design for adaptive support in the form of strategic prompts are effective for manipulating student behavior in productive ways and for supporting learning. These results also motivate specific modifications to the original design.