This study examines how experts and novices differ in their epistemic and dialogic engagement with generative artificial intelligence (GenAI) during collaborative problem solving (CPS), from a Knowledge Building perspective that conceptualizes CPS as collective knowledge creation. Four pairs of undergraduate students and one pair of experts (science education researchers) worked with ChatGPT-4o to address socio-scientific issues related to food breeding and genetic modification. Their interactions were analyzed using Epistemic Network Analysis, based on seven dialogic moves: problem definition, new ideas, promisingness evaluation, comparison, critical discourse, meta-dialogue, and higher-level ideas. Results showed distinct discourse patterns across groups. Experts exhibited broad and integrated epistemic networks, enabling sustained reorganization of inquiry and shaping prompts that elicited knowledge-building responses from GenAI. In contrast, novice interactions were primarily organized around localized connections between idea generation and evaluation, resulting in more reactive and fragmented engagement with AI outputs. Findings suggest that GenAI should be conceptualized not as an autonomous collaborator but as a responsive epistemic resource whose role is constituted through learners’ dialogic practices. From this perspective, the educational value of GenAI-supported CPS lies not in the sophistication of AI outputs but in how epistemic framing and dialogic engagement are scaffolded within collaborative inquiry.

A challenge in interdisciplinary learning arises from the disciplinary boundaries brought to the collaboration contexts. Studies have designed technology- and artificial intelligence (AI)-mediated interdisciplinary learning environments to break disciplinary boundaries. However, these studies did not fully address instructional design for students with high-level disciplinary backgrounds. To address this challenge, we highlighted the role of generative AI (GenAI) as a boundary broker. We designed an instructional approach that enabled boundary crossing using GenAI in interdisciplinary research development and investigated its impacts on collaboration. We designed a workshop in which 14 doctoral students from various fields formed interdisciplinary teams to develop research proposals collaboratively. Firstly, they participated in exploratory idea-exchange activities using GenAI to identify potential collaborators. They then received further support from GenAI for developing interdisciplinary research proposals. We analyzed the discourse and AI usage records of two high-performing teams to understand how GenAI was used in interdisciplinary research development, contributed to their idea improvement, and affected their boundary-crossing patterns. The results indicate that GenAI (1) provides adaptive support at both the group and individual levels, (2) supports interdisciplinary discourse by offering potential boundary objects and promotes the emergence of epistemic agency, and (3) transforms boundary-crossing patterns, enabling individuals to leverage their expertise in interdisciplinary collaboration more effectively. We conclude that GenAI contributed to students breaking their disciplinary boundaries by providing a boundary object. Our design and analysis methods can be applied to other GenAI-supported collaborative learning contexts in which learners have different levels of knowledge, skills, and attitudes.

ABSTRACT Background Research on collaborative learning has long examined processes and outcomes separately, with distinct fields—Computer‐Supported Collaborative Learning (CSCL), learning sciences and recently learning analytics—drawing on differing epistemological premises. Integrated analytical accounts connecting collaboration processes to the outcomes of collaboration remain scarce. This special issue addresses that gap by studies of collaborative learning that employ conceptualisations and analytical methods aligned with a transactional approach. Contributions in This Special Issue The issue comprises an editorial, seven empirical studies and two commentaries. Collectively, the studies show that scaffolding collaboration can target reflection, regulation, motivation, knowledge‐object development and learner–technology relations, each mediating transactions differently. Using multimodal and multilayered analytical approaches—including epistemic network analysis, multi‐layered temporal network analysis, sequence mining and multilevel modelling—the contributions examine how tools, artefacts, discourse and motivational states jointly shape learning outcomes across authentic educational settings, from higher education to school and after‐school contexts. The commentaries situate these findings within broader theoretical and methodological debates in CSCL and research that emphasises understanding the relational and situated nature of collaborative learning. Implications for Future Research The authenticity of the studied settings carries implications for data collection, analysis and design for collaborative learning, offering insight for researchers, teachers, learning designers and technology developers. Future research should give attention to the relational, temporal, multimodal aspects combined—that is, employ an integrative approach in the interpretation and revisit conceptualisations that integrate material‐ecological, epistemic‐discursive and motivational‐psychological perspectives on the relationship between collaboration processes and outcomes.

BackgroundThis study aimed to develop and test new analytics for knowledge-building practices from the transactive perspective. Based on a literature review, network analysis was identified as a promising analytical tool for these practices. We observed two aspects of network analysis that could be further developed: the multilayers of networks and temporality.ObjectivesConsequently, the study investigated transactive processes of collaborative learning leading to different learning performance levels using the multi-layered temporal network analysis after examining the advantages of the multi-layered temporal network analysis by comparing its findings with those of the traditional discourse network analysis.MethodsThis method was applied to identify multi-layered temporal discourse patterns in knowledge-building practices among first-year university students engaged in project-based learning. Discourse in each group was decomposed into discourse topics using exploratory clustering analysis with temporal changes in all nouns' degree centralities. Then, the multi-layer discourse patterns were compared between groups with different learning performance levels.Results and ConclusionsWe identified two conditions for high learning performance not found by the traditional network discourse analysis: extensive comparison of multiple ideas and co-elaboration through warranting. For idea improvement in knowledge-building practices, the judgement of idea promisingness is crucial. Groups with high learning performance engaged in this judgement by contrasting multiple ideas, a strategy not found in groups with low learning performance. Further, of the two dimensions of idea improvement, the co-elaboration process was evident in learners' discourse around their promising ideas, facilitated by warranting. Thus, the multi-layered temporal network analysis of discourse could provide more detailed descriptions of how learners engage in their idea improvement processes. Comparative case studies suggest hypothetical conditions for successful learning processes.

Since the inception of learning analytics, the CSCL community has been applying a range of methods to examine collaborative discourse, such as natural language processing and networked approaches. Despite significant advancements in methods, a critical gap persists in translating these data-driven insights into actionable strategies to inform pedagogical decisions. The proposed hybrid symposium seeks to bridge this gap by engaging learning scientists, CSCL researchers, and educators in discussions that integrate advanced discourse analytics with pedagogical design. Through five diverse presentations, the symposium will explore the integration of advanced discourse analysis techniques, such as networked approaches, multimodal analytics, and clustering analysis, into pedagogically meaningful designs and actionable pedagogical insights for collaborative learning environments. By fostering dialogue on the challenges and opportunities in contextualizing computational insights, this symposium aims to establish actionable approaches for enriching collaborative discourse and advancing the broader CSCL community’s theoretical, methodological, and practical understanding.

This article explores how non-native English-speaking (NNES) students and higher education educators perceive AI-generated feedback through dialogic feedback framework and ethical lens. Using a quantitative ethnography approach, Study 1 employed focus group interviews with 17 NNES students and nine educators, identified three themes: (1) instrumental efficiency vs. holistic understanding, (2) emotional comfort and social dynamics , and (3) ethical issues for trustable feedback. Study 2 employed epistemic network analysis (ENA) to identify that educators' epistemic frames integrated broader ethical considerations, while students' frames emphasized emotional and relational aspects to human educators. Findings indicate that while AI-generated feedback offers emotional neutrality and immediacy, it lacks contextual depth, relational continuity, and interpretive richness in human feedback. This suggests that AI-generated feedback should be designed to work alongside, not replace human educators' feedback, offering implications for responsible and dialogically informed feedback practices in higher education.

This design-based research investigated the differences in knowledge-creation practices among university students across two blended learning formats—traditional and hybrid—over multiple years, guided by knowledge-building principles. Each year, 74 first-year students participated in the course to develop new happiness indices through small-group activities. They used a Computer-Supported Collaborative Learning (CSCL) system to share weekly reflection notes and plan future activities in addition to face-to-face interactions. In the hybrid blended learning format, students had to manage communication with remote group members. The students’ reflection notes in CSCL were analyzed to evaluate the development of their epistemic views and perceptions of the community of inquiry (CoI). Clustering analysis revealed that in the traditional blended learning year, 57

Successfully engaging in collaborative problem solving is an essential constituent of professional and general life skills.Collaborative problem solving is also used to engage learners in collaborative processes relevant to the acquisition of knowledge and skills.Here, learners co-construct knowledge by transactively building on each other's contributions.In this symposium, we explore the factors leading to successful engagement in collaborative problem solving.The conceptualization of transactivity in collaborative problem solving is refined balancing the merits of both sharing new knowledge and referring to others' contributions.Empirical studies in this symposium show that it is important for successful collaborative problem solving that individuals elicit and share information with their learning partners who should evenly participate.Further, engaging in relevant collaborative processes can be facilitated by scaffolds such as sentence openers facilitating learners referring to each other. Symposium introductionCollaborative problem solving is meant to engage learners in joint discourse and knowledge co-construction.CSCL research particularly inquires under which circumstances collaborative problem solving facilitates individual and joint learning (Roschelle & Teasley, 1995).Here, the most promising collaborative learning integrates both solving socio-cognitive conflicts and mutual support towards deeper understanding.Collaboratively exchanging and scrutinizing each other's ideas, arguments, and reasoning towards the problem solution is key to the co-construction of knowledge.Transactivity (i.e., learners building on each other's reasoning) has been identified as a relevant quality predicting collaborative learning outcomes.Yet, the rather general conceptualization of transactivity allows a range of different operationalizations, partly leading to inconsistent results.Further aspects in collaborative problem solving need to be considered interacting with CSCL 2023 Proceedings © ISLS 338 transactive discourse and how transactivity facilitates learning.These aspects concern how activities and engagement are distributed in the group's discourse.Another aspect is how learners share information during collaborative problem solving.This highly depends on knowledge and information accessible to the learners such as prior knowledge or other resources.The emergence of these aspects in collaborative problem solving and their relevance for knowledge co-construction is being investigated in the papers collected in this symposium.In the first paper, Vogel & Weinberger conceptualize transactivity as relevant mechanism for successful collaborative problem solving.It extends the commonly used definition by balancing the relevance of both adding novel information and building on other's contribution in collaborative discourse.The empirical studies in this symposium shed light on the above-mentioned aspects of successful collaborative problem solving and how they may be related to transactivity.The study by Hong et al. explores collaboration patterns in problem-based learning, revealing that the individual contributions in collaborative discourse are distributed less equally the more active the groups are.With more activity being related to better learning outcomes, this raises the question how equal engagement can be supported in problem-based learning.To this effect, Oshima et al. analyzed transactive discourse in collaborative problem solving using temporal network analysis.They found that the distribution of contributions in transactive discourse may not be predicted by introducing fixed or alternating leadership roles but rather by the access to and use of artifacts relevant to the problem to be solved.The paper by Brandl, Richters, et al. sheds more light on the relevance of contributing knowledge and information in collaborative problem solving.In their study about interdisciplinary problem solving in the context of collaborative diagnostic reasoning, successful learning processes were indicated by the learners' conceptual knowledge enabling them to share an adequate initial problem representation.Finally, the study by Rejon et al. compared experimentally different types of sentence starters for novelty and reference to support transactive discourse in collaborative problem solving.The analyses show that sentence starters inducing reference were linked with more engagement in sociodiscursive activities expected to facilitate the co-construction of knowledge in the collaborative learning process.

Collaborative learning fosters the ability to creatively solve problems in collaboration with other learners. Researchers in learning science have transcribed learners' speech to qualitatively analyze collaborative learning to reveal various patterns that increase learning performance. Although prior studies have identified speakers to support the process of transcription, those studies were limited in simultaneously identifying multiple speakers. We propose a novel speaker-identification algorithm that can simultaneously recognize multiple speakers using business-card-type sensors. The algorithm can remove ambient noise with low-cost sensors and still identify multiple simultaneous speakers. The experimental evaluations show that the algorithm accurately identifies simultaneous multiple speakers in a multi-person activity under conditions with varying numbers of users, environmental noise, and users' short utterances.

We examined a course designed for pre-service teachers to develop their epistemic cognition on teaching practices through collaboratively learning pedagogical knowledge.Preservice teachers' epistemic cognition was evaluated by asking them to interpret a teaching practice recorded in a video, before and after the course.Three patterns of epistemic cognition were identified using clustering analysis.Characteristics of each pattern were discussed for further refinement of the course design.

Collaborative learning has been qualitatively analyzed by learning science researchers to enhance learning performance. A quantitative analysis system supports the existing qualitative analysis of collaborative learning. We propose an Internet of Things (IoT) system comprising business-card-type badges, radio-over-fiber (RoF)-based synchronization, and a collaboration analysis algorithm to support collaborative learning analytics in different venues (e.g., offline, online, and hybrid). We showed that our proposed system quantitatively supports qualitative analysis of collaborative learning in different environments.

Transactivity in discourse is key to successful learning.In this study, we propose a new temporal network analysis to visualize and calculate how small groups of learners engage in transactive discourse.Our temporal network analysis identified three leadership patterns in transactive discourse: collective, rotating, and fixed.Further interaction analysis revealed critical differences in the division of labor between a fixed and rotating leadership group.

This study examines the design activities of engineers and product designers from the perspective of knowledge building. The practice of knowledge building has been studied for more than 30 years. However, in recent years, analytical methods have been developed to analyze it from two directions—idea improvement and epistemic frames—and these methods are currently being enhanced. Nevertheless, studies that have analyzed idea improvement and epistemic frames have focused on practices in the classroom rather than discussing the activities of engineers and designers, who are also knowledge building models. Therefore, this study analyzed the co-design process of a product design team and an engineering team that engaged in creative activities for their work from the perspectives of idea improvement using socio-semantic network analysis (SSNA) and the epistemic frame by epistemic network analysis (ENA). Moreover, this study discussed defining meaning segments using SSNA as a computational approach for quantitative ethnography (QE). As a result, both teams showed good knowledge building characteristics in that they continuously improved their ideas. Furthermore, the engineering team worked under various epistemic actions, while the product designers worked under a limited epistemic frame. We also confirmed that the analysis method of this study was able to extract the characteristic discourse of each team. These results support future knowledge building practices, as they illustrate that designers and engineers engage in the same continuous idea improvement under different epistemic actions. Furthermore, this study contributes to future QE research because the results show the qualitative differences between designers and engineers using determining meaning segments as a computational approach.

Collaborative learning is an educational approach to teaching and learning that involves groups of learners collaborating to solve a problem, complete a task, or create a product. To enhance the performance of collaborative learning, the studies in Yamaguchi et al. (2021, 2021, 2021, and 2022) develop an IoT system and quantitatively extract collaboration between learners. The studies acquire sensor data from IoT badges on learners and analyze learning activities with the acquired sensor data on a computer. However, existing studies are not user-friendly for learning analysts who are unfamiliar with information technology owing to complex software installation and command line interface (CLI) operation. Such drawbacks hinder the wide expansion of technology and the exploration of new learning patterns in learning science. Considering high usability for analysts, this paper proposes novel web services named Sensor-based Regulation Profiler Web Services (SRP Web Services) for collaboration analysis with IoT badges. The proposed web application consists of front-end on Next.js and back-end on FastAPI, SQLite, and Python and extracts key points in learning activities for the analysts from the acquired sensor data on a web browser. Experimental evaluations showed that the proposed web services support learning analysts in quantitative analysis of learning activities with high usability. In addition, SRP Web Services are scalable with hundreds of users.

Despite the promise of quantitative ethnographic approaches for visualizing the trajectories of change over time (temporal analysis) further work is needed to develop strategies for accurately representing phenomena. This holds especially true for identifying the relational context of discourse, which includes the creation of time units that group lines of data for the purpose of interpretation. While in-depth interpretive review of discourse may serve as the ‘gold standard’ for identification of thematic time units, this approach is tedious and may not be appropriate for larger datasets. Incremental approaches, such as creating a new unit for every ten lines of chronological data, are functional for larger datasets, but may lack nuance. This work introduces the Knowledge Building Discourse Explorer (KBDeX), which computationally identifies relational units using socio-semantic network analysis, allowing for the identification of time units based on characteristics of the discourse that can be systematically applied to larger datasets. To examine the utility of each approach, epistemic networks of COVID-19 press releases from seven countries were created with time units derived from the incremental and computational approaches, which were then compared to the interpretive approach. Results indicated that KBDeX and incremental network means were closer to the ‘gold standard’ interpretive approach in some instances. Two countries’ trajectories are examined in greater depth to understand when each approach might be most appropriate. The work concludes with a discussion of the affordances and constraints of each approach, and contexts in which they may be useful.

Collaborative learning practices foster the ability to solve creative problems in collaboration with other learners. The collaboration enables learners to learn new ideas from other learners and enhances the social ability of the learners through interaction with other learners. Although the learning science field now uses qualitative analysis to analyze the effects of the collaborative discourse, qualitative analysis requires much human and time costs to analyze the collaborative discourse with dozens of students. This study proposes Sensor-based Regulation Profiler to reduce the analysis costs. The proposed scheme consists of the business card-type sensors that acquire sensor data from each learner with a precise time synchronization as well as learning analysis methods that analyze the collaborative discourse from the acquired sensor data. Experimental evaluations using the proposed scheme showed that the proposed business card-type sensors realized a time synchronization error of 7.7µs on average across the sensors. In addition, the proposed learning analysis could extract and visualize the collaborative activity of each learner in the collaborative discourse through the social graph extraction, learning phase extraction, speaker identification, and activity estimation by using the sensor data from the proposed business card-type sensors.

Artificial intelligence (AI) and new technologies are having a pervasive impact on modern societies and communities. Given the potential of these new technologies to transform the way things are done, it is important to understand how they can be used to support inclusive education, particularly regarding minority students. This systematic review analyzes the advantages and challenges of using AI and new technologies in different sociocultural contexts, and their impact on minority students. In terms of advantages, this review found that AI and new technologies (a) improved student performance, (b) encouraged student interest in STEM/STEAM, (c) promoted student engagement, and (d) showed other advantages. This review also identifies the main challenges associated with the use of AI and new technologies for inclusive education: (a) technological challenges, (b) pedagogical challenges, (c) dataset limitations, (d) low satisfaction using technology, and (e) cultural differences. This review proposes some solutions to these challenges at the pedagogical, technological, and sociocultural levels, and also explores important aspects of inclusive education that address the students’ sociocultural diversity. The findings and implications will aid teachers, practitioners, and policymakers in making decisions on the effective use of AI and new technologies to support sociocultural inclusiveness in education.