Wasserstein-Fisher-Rao Embedding: Logical Query Embeddings with Local Comparison and Global Transport
Lists
Only you see this
Abstract
Answering complex queries on knowledge graphs is important but particularly challenging because of the data incompleteness. Query embedding methods address this issue by learning-based models and simulating logical reasoning with set operators. Previous works focus on specific forms of embeddings, but scoring functions between embeddings are underexplored. In contrast to existing scoring functions motivated by \textit{local} comparison or \textit{global} transport, this work investigates the \textit{local} and \textit{global} trade-off with unbalanced optimal transport theory. Specifically, we embed sets as bounded measures in $\real$ endowed with a scoring function motivated by the Wasserstein-Fisher-Rao metric. Such a design also facilitates closed-form set operators in the embedding space. Moreover, we introduce a convolution-based algorithm for linear time computation and a block-diagonal kernel to enforce the trade-off. Results show that WFRE can outperform existing query embedding methods on standard datasets, evaluation sets with combinatorially complex queries, and hierarchical knowledge graphs. Ablation study shows that finding a better \textit{local} and \textit{global} trade-off is essential for performance improvement.\footnote{Our implementation can be found at~\url{https://github.com/HKUST-KnowComp/WFRE}.}
BibTeX
@inproceedings{wang2023wasserstein,
author = {Zihao Wang and Fei Weizhi and Hang Yin and Yangqiu Song and Ginny Wong and Simon See},
title = {{Wasserstein-Fisher-Rao Embedding: Logical Query Embeddings with Local Comparison and Global Transport}},
booktitle = {Annual Meeting of the Association for Computational Linguistics},
year = {2023},
doi = {10.18653/v1/2023.findings-acl.864},
}