About the semantic-specific embedding

[ZhaozwTD]

After formula five, you said "we can obtain P groups of semantic-specific node embeddings" . I looked through the full paper and found no expression or formula for this semantic-specific node embeddings. Is it possible to concatenate the learned embedding of node i for the meta-path Φ in formula four?