Correction to: Many are the ways to learn: Identifying multi-modal behavioral profiles of collaborative learning in constructivist activities

Abstract

In this article the title was incorrectly given as 'Many are the ways to learn identifying multi-modal behavioral profiles of collaborative learning in constructivist activities' but should have been 'Many are the ways to learn: Identifying multi-modal behavioral profiles of collaborative learning in constructivist activities'.Similarly, a few citations of our previous work have been incorrectly placed:-On page 12, in section Dataset and Preprocessing paragraph 5 ("This dataset, with multi-modal behaviors as well as…."), the citation '(Nasir, Bruno, et al., 2021a) should have been 'Nasir, J., Norman, U., Bruno, B., Chetouani, M., & Dillenbourg, P. (2021b)'. -On page 12, in section Dataset and Preprocessing paragraph 5 ("This dataset, with multi-modal behaviors as well as…."), the citations 'Nasir, Bruno, et al. (2020; 2021)' should have been 'Nasir, Bruno, et al. (2020a; 2021a)'. -On page 14, in subsection Analysis Approach paragraph 1 ("The goal of this work is to build and understand…"), the citations '(Nasir et al., 2021a, b; Nasir, Norman, Bruno, & Dillenbourg, 2020c)' should have been '(Nasir et al., 2021a; 2020a)'. -On page 15, in Fig. 4, the citation 'Nasir, Norman, Bruno, and Dillenbourg (2020c)' should have been 'Nasir, J., Bruno, B., & Dillenbourg, P. (2020a)'.

Comparing the psychometric properties of two primary school Computational Thinking (CT) assessments for grades 3 and 4: The Beginners' CT test (BCTt) and the competent CT test (cCTt)

“…As numerous researchers have put forward, instruments such as the BCTt and the cCTt should be combined with other forms of assessments in a systems of assessments (Grover et al, 2015 ; Román-González et al, 2019 ; Weintrop et al, 2021a ) to accurately measure the full range of competencies at play when considering CT (Brennan and Resnick, 2012 ; Piatti et al, 2022 ). The systems of assessments could therefore include other instruments which assess CT practices such as the test by Li et al ( 2021 ), employ direct observations of students' thought processes and strategies (Lye and Koh, 2014 ; Chevalier et al, 2020 ), or learning analytics and educational data mining techniques (Cock et al, 2021 ; Nasir et al, 2021 ; Zapata-Cáceres and Martín-Barroso, 2021 ). Complementary assessments would not only help gain a more accurate and in-depth picture of student learning but also feed into the learning activity design and intervention process (Chevalier et al, 2022 ).…”

Personalized Productive Engagement Recognition in Robot-Mediated Collaborative Learning

“…In addition to this, it also contains the performance metrics and the learning gains of the teams. We must note here that while the dataset provides 5048 data points with 34 teams; the learner profles (discussed later in Section 4.1) are only available for 32 teams [27]; hence, the data used in this work consists of 4676 data points from the dataset.…”

“…Productive Engagement Scores -For the Productive Engagement Score, we make use of the metric proposed by Nasir et al [24] to characterize productive engagement in real-time. Briefy, the Figure 1: Overview of the proposed approach: Learners are clustered into three profles based on their multimodal behavioural features [26] and learning gain: Expressive Explorers (Type 1 gainers), Calm Tinkerers (Type 2 gainers), and Silent Wanderers (non-gainers) [27]. Behavioural features and learners' profles are then fed as inputs to the Efcient Neural Architecture Search (ENAS) framework to automatically search for the optimal architecture for each profle and for each modality to predict productive engagement.…”

“…The proposed concept was validated in the context of a robot-mediated human-human collaborative learning activity implemented within the JUSThink platform by Nasir et al [28], where the employed approach surfaces multiple ways (precisely sets of multimodal behaviors given by log, audio, and video data) in which diferent individuals learn [25]. Through an in-depth quantitative and qualitative analysis, three learner profles are identifed, namely, Expressive Explorers, Calm Tinkerers, and Silent Wanderers where the frst two profles correspond to those who have high learning gain, while the last one corresponds to those who exhibit a low learning gain [27]. The idea of conceptualizing engagement with a focus on the learning aspect in educational settings and the aforementioned outcomes indeed show potential in advancing the literature in educational HRI and multimodal learning analytics.…”

“…We extend the approach proposed by Salam et al [39] to the task of productive engagement prediction. Hence, we propose to learn personalized deep architectures for diferent learner profles identifed by Nasir et al [27] using single-task and multi-task Efcient Neural Architecture Search (ENAS). Via multitask ENAS, we investigate whether simultaneously predicting productive engagement scores and classifying learners profles can efciently ofer an implicit personalization of productive engagement models.…”