AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

CLUHCS:Dual-View Contrastive Learning Enabled Unsupervised Heterogeneous Community Search with Meta-Path Behavior Modeling

Xiaoqin Xie, Bin Zhao, Mingzhu Chang, Shuai Han, Wu Yang

AAAI 2026 · Graph learning › Graph self-supervised learning · approach: Contrastive learning with negatives and views

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use PathSim-selected positive samples instead of all meta-path neighbors (CLUHCS-A) (ablation)CLUHCS · ACM and DBLPprecision, recall, F1, JaccardImproved
use PathSim-selected positive samples instead of all meta-path neighbors (CLUHCS-A) (ablation)CLUHCS · IMDBprecision, recall, F1, JaccardNo clear effect
add positive sampling from meta-path neighbors (vs none, CLUHCS-Z) (ablation)CLUHCS · ACM, DBLP, IMDBall metricsImproved
add the meta-path view (vs CLUHCS-MP) (ablation)CLUHCS · ACM, DBLP, IMDBF1, JaccardImproved
add the relation view (vs CLUHCS-Rel) (ablation)CLUHCS · ACM, DBLP, IMDBF1, JaccardImproved
use ESG greedy search instead of clustering or top-k search over fixed embeddings (ablation)CLUHCS search layer · fixed embeddingscommunity qualityImproved
vary hop count K (ablation)CLUHCS · ACM, DBLP, IMDBF1, Jaccard: optimal K=6 (ACM), K=4 (DBLP), K=7 (IMDB)Mixed
raise PathSim sampling threshold mu (ablation)CLUHCS · ACM vs DBLP and IMDBF1, Jaccard: ACM best at mu=0.8; DBLP and IMDB best at mu=0.1Mixed
raise cohesion parameter omega (ablation)CLUHCS · ACM, DBLP, IMDBprecision, recall, F1: DBLP peaks at omega=0.5Mixed
use CLUHCS instead of supervised FCS-HGNN variantscommunity search · DBLPprecisionImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy