Analytics of Behavior Semantics for Understanding Constraint Conditions Hidden in Formative Process of Real-world Learning

Abstract

An alternative of classroom learning is situated learning by behaving in the world (e.g., environmental learning in a natural field). Among various types of human intelligence, this research is interested in understanding the process mechanism that human intelligence is formed through learner-learner and learner-environment interactions. Here, we assume that a learner’s cognition, interpretations and behavior in the world are positively or negatively affected by various levels of constraint conditions given by his/her body, cognition, and surroundings. For example, a learner may not generate a certain type of effective real-world behavior if he/she does not know basic knowledge (i.e., cognitive-level constraint). At the place where interesting objects do not exist in the world, a learner’s active inquiry will be restricted (i.e., environment-level constraint). To mine a learner’s prospective behavior for making multi-view understanding of the world, we developed technologies (1) to multidirectionally sense a learner’s behavior in the world, (2) to parameterize time-series behavior with different semantics, and (3) to extract constraint conditions hidden in the formative process of real-world learning. This research applied our analytical framework to our experiments of environmental learning with 30 participants in an experimental forest. Our initial results showed that the semantic-level data of behavior enabled us to understand the cognitive state and constraints of learners, and to find the changing points of learning situation. This illustrates our framework can be a theoretical basis to understand the mechanism of situated intelligence emerging in the world.



Author Information
Koryu Nagata, Shizuoka University, Japan
Masahiro Tada, Kindai University, Japan
Masaya Okada, Kyushu University, Japan

Paper Information
Conference: ACE2020
Stream: Assessment Theories & Methodologies

This paper is part of the ACE2020 Conference Proceedings (View)
Full Paper
View / Download the full paper in a new tab/window


Posted by amp21