arXiv ScienceSearch

arXiv subjects

Baiyue He

Publications and source records attributed to Baiyue He.

2 recordsLinked to original sources

Heterogeneous Interaction Network Analysis (HINA): A New Learning Analytics Approach for Modelling, Analyzing, and Visualizing Complex Interactions in Learning Processes

Existing learning analytics approaches, which often model learning processes as sequences of learner actions or homogeneous relationships, are limited in capturing the distributed, multi-typed interactions in contemporary learning environments. To address this, we propose Heterogeneous Interaction Network Analysis (HINA), a multi-level learning analytics framework for modelling interactions and associations across diverse entities (e.g., learners, behaviours, and AI agents) in learning processes. Grounded in network science principles, HINA integrates a multi-level analytical framework that analyzes individual process metrics (node-level), prominent associations (dyad-level), and latent clusters (meso-level) to address questions about how different elements in a learning environment interact and co-influence each other. In this paper, we first detail the theoretical and mathematical foundations of HINA for individual, dyadic, and meso-level analysis. We then demonstrate HINA's utility through a case study on AI-assisted small-group collaborative learning, revealing students' interaction profiles with peers versus AI, distinct engagement patterns that emerge from these interactions, and specific types of learning behaviors (e.g., asking questions) directed to AI versus peers. HINA contributes a novel and unified analytical framework that enables researchers to quantify individual-level processes, identify significant associations, and uncover meaningful clusters within a single workflow. Accompanied by a dedicated web tool, HINA provides a complete workflow for supporting process-based assessment, and enables a new theoretical lens for modeling and understanding complex and mediated learning processes.

cs.SI

Networks of amenities reveal universal homophily and heterophily across global cities

Agglomeration economies drive urban growth at different spatial scales by enabling productivity gains, knowledge spillovers, and shared inputs among proximate firms and amenities. To develop a unified science of cities it is thus important to understand how and to what extent different amenities cluster or mix across scales and regional contexts. By utilizing a novel Bayesian framework for nonparametrically quantifying the spectrum of possible mixing patterns of amenities in a city, we identify universal spatial scales of homophily (agglomeration) and heterophily (co-agglomeration) among different amenity types across roughly 800 cities worldwide. Through a detailed longitudinal case study, we also find that the changes in heterophilic mixing derived from our methodology more effectively predict changes in neighborhood rental values than the diversity of amenities present. These findings suggest that agglomeration economies exhibit universal spatial regularities that depend largely on the types of firms or amenities being considered, rather than their specifics or regional context, and highlight the benefit of heterophilic amenity mixing at walkable spatial scales.

physics.soc-ph