A recommendation-based directed-graph framework for construction activity sequencing

doi:10.1016/j.asoc.2026.115937

A recommendation-based directed-graph framework for construction activity sequencing

2026-07-08

0

PRE

AI

DOI：10.1016/j.asoc.2026.115937

Original

Original request for help

Save

Abstract

En 中文

• GNNs fail at short activity flow; classification misfits sparse top-K sequence task. • A novel graph recommender ranks next activities via embedding-based sequencing. • Developed ARM, DeepWalk and node2vec outperform GCN, GAT, GraphSAGE under data sparsity. • Latent factors enable generalization across projects without manual features. • Recommender model predictions converted into editable activity sequencing graphs. .

A recommendation-based directed-graph framework for construction activity sequencing

Abstract

Journal

Organization