DBP15K_JA_EN

class DBP15K_JA_EN(root: str | Path, download: bool | None = False, train_ratio: float | None = 0.2, dtype: dtype = torch.float32, seed: int | None = 0)

Bases: Dataset

A pair of Japanese to English version of multi-lingual DBpedia networks. The dataset is proposed by the “Cross-lingual Entity Alignment via Joint Attribute-Preserving Embedding” paper, and the node attributes are given by pre-trained and aligned monolingual word embeddings from the “Cross-lingual Knowledge Graph Alignment via Graph Matching Neural Network” paper. There are 15,000 pairs of aligned entities in DBP15K (Japanese to English).

Graph	#nodes	#edges	#node attrs	#edge attrs
JA	19,814	77,214	300	0
EN	19,780	93,484	300	0