Meituan Tech Salon 57 – Knowledge Graph Practice (CIKM2020) Special Session

The Meituan Tech Salon (Session 57) is an online event held on November 21, 2020, jointly organized by Meituan's technical team and scientific association. It features academic talks related to the CIKM2020 conference, focusing on knowledge graph and AI applications.

Talk 1 – Optimizing Recommendation Systems with Multi‑Modal Knowledge Graphs (Speaker: Cao Xuezhi, Meituan Technical Expert). The presentation introduces the Multi‑Modal Knowledge Graph Attention Network (MKGAT) to address data sparsity and cold‑start issues by leveraging textual and visual modalities.

Talk 2 – Graph Embedding for Personalized Search (Speaker: Gu Wanli, Meituan Algorithm Engineer). It proposes GraphLSR, a graph‑embedding based logical structure representation learning method that models user‑query‑item interactions using three key graph patterns.

Talk 3 – Query‑Aware Generation of Recommendation Reasons (Speaker: Yang Yang, Meituan Algorithm Engineer). The work presents a query‑sensitive generation framework that incorporates the user query into both encoder and decoder to produce personalized recommendation explanations.

Talk 4 – Self‑Supervised Sequential Recommendation via Mutual Information Maximization (Speaker: Zhou Kun, Meituan Algorithm Intern). The paper introduces S³‑Rec, a self‑attention network trained with four self‑supervised tasks and mutual information maximization to improve sequential recommendation under sparse data.

All talks are available for viewing after the event, and registration details are provided in the original announcement.