The article explains how legacy monitoring based on isolated metrics, traces, and logs cannot keep up with the massive, fragmented, and dynamic data of modern IT systems, and introduces UModel—a graph‑based observability model that bridges data, model, and engineering gaps to enable AI‑driven operations.
This article examines the evolution of IT observability, explains why traditional metrics, traces, and logs fall short for AI‑driven operations, and introduces UModel—a graph‑based universal observability model that structures fragmented data into a semantic runtime context for autonomous AIOps agents.
FantasyHSI introduces a graph‑based multi‑agent framework that combines visual‑language models and video‑generation diffusion to let digital humans perceive, plan, and interact autonomously in any 3D scene, producing physically plausible, long‑duration actions for animation creation and embodied‑AI simulation.
This presentation reviews the background, significance, current research status, objectives, and key contributions of a graph‑based approach to information extraction, covering entity recognition, relation extraction, event extraction, open‑domain extraction, and the proposed unified modeling framework with experimental results.
The talk presents an overview of graph neural network development, explains 极验's graph modeling research and evolution, and details offline and real‑time GCN solutions, including self‑supervised training, large‑scale handling, and performance comparisons, highlighting practical applications in fraud detection and risk control.
The winning KDD Cup 2020 debiasing solution builds a heterogeneous item‑to‑item graph with click‑co‑occurrence and multimodal similarity edges, uses multi‑hop random walks to generate unbiased candidate samples, trains LightGBM with a popularity‑weighted loss, and aggregates scores to lift low‑popularity items, thereby eliminating selection and popularity bias and achieving first place among 1,895 teams.
This talk reveals how Youku built a flexible, universal user‑reach platform that leverages graph‑based task modeling, dynamic expressions, and a unified execution engine to deliver recall and activation campaigns to over a billion users with fast, precise targeting and experiment‑driven optimization.
