Scholarlink

This article discusses the issues of embedding for a metagraph model. The Type I and Type II complex graph descriptions are given. A formal definition of the metagraph model is given. The basic concept of a metagraph model is a metavertex, which can include nested vertices, metavertices, and edges. The general framework of a metagraph embedding is proposed, based on the concepts of embedding complex graphs of Type I and Type II, as well as on the transformation of complex graphs of Type II into Type I. An alternative definition of the metagraph model is proposed, which is based on the concepts of red and blue graphs. The metagraph embedding approach based on alternative definition is proposed. DeepWalk algorithm, TransE method, and TransR method were used for red and blue graphs embeddings. The embedding results are evaluated using Hits@k metric. Thus, embedding based on an alternative definition of the metagraph model can be considered as a promising direction for metagraph embedding.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Metagraph Embedding Approach Based on Red-Blue Graphs

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Stepan Vinnikov,
  • Anatoly Nardid,
  • Yuriy Gapanyuk

摘要

This article discusses the issues of embedding for a metagraph model. The Type I and Type II complex graph descriptions are given. A formal definition of the metagraph model is given. The basic concept of a metagraph model is a metavertex, which can include nested vertices, metavertices, and edges. The general framework of a metagraph embedding is proposed, based on the concepts of embedding complex graphs of Type I and Type II, as well as on the transformation of complex graphs of Type II into Type I. An alternative definition of the metagraph model is proposed, which is based on the concepts of red and blue graphs. The metagraph embedding approach based on alternative definition is proposed. DeepWalk algorithm, TransE method, and TransR method were used for red and blue graphs embeddings. The embedding results are evaluated using Hits@k metric. Thus, embedding based on an alternative definition of the metagraph model can be considered as a promising direction for metagraph embedding.