Dual Analysis of Literature and Existing Tools for Establishment of Acceptable xAPI Profile for Collaborative Learning

Abstract

The increasing adoption of collaborative learning in digital education has amplified the need for standardized, interoperable methods to capture and analyze learner interactions. While xAPI (Experience API) provides a flexible syntax for logging learning activities, its effectiveness in supporting cross-platform analytics depends on domain-specific xAPI Profiles that define shared vocabularies and structural patterns. However, no widely accepted profile exists for collaborative learning, which involves complex, socially situated interactions that are challenging to formalize and compare across platforms. This study aims to design and evaluate an xAPI Profile for collaborative learning that is grounded in both theoretical and practical evidence. To achieve this, we conducted a two-pronged analysis. First, a systematic review of 72 peer-reviewed research articles in Computer-Supported Collaborative Learning (CSCL) was carried out to extract commonly studied interaction categories, such as proposing, agreeing, disagreeing, and summarizing. Second, we analyzed anonymized interaction logs from widely used educational tools—including LoiLoNote School, Miraiseed, and schoolTakt—used in Japanese elementary and junior high schools, to identify frequently occurring learner actions and behavioral patterns. Based on the findings, we developed an xAPI Profile by the ADL xAPI Profile specification, which includes well-defined verbs, activity types, statement templates, and behavioral patterns. The proposed profile was evaluated for representational coverage and semantic alignment by comparing it against both the literature-derived categories and tool-based log data. The results show that the profile covers a substantial proportion of both theoretically important and practically observed interaction types, while also highlighting meaningful gaps between research models and real-world practice. This study contributes to the field of learning analytics by providing a replicable methodology for empirically grounded xAPI Profile design, and by offering a reusable data model that facilitates cross-platform analysis of collaborative learning. The profile has the potential to support interoperable data collection, visualization, and feedback systems, thereby advancing both research and practice in collaborative learning analytics.