Rethinking Identity in Knowledge Graph Embedding

Jiayi Li, Jiaqi Sun, Yujiu Yang

Keywords: Knowledge Graph Embedding, Knowledge Graph Completion, Bilinear Based Models

TL;DR: We scrutinize the identity relation in knowledge graphs, find that bilinear based models fail to uniquely model it, and propose a solution with other good properties.

Abstract:

Knowledge Graph Embedding (KGE) is a common method to complete real-world Knowledge Graphs (KGs) by learning the embeddings of entities and relations.
Beyond specific KGE models, previous work proposes a general framework based on group. A group has a special element identity that uniquely corresponds to the relation identity in KGs, which implies that identity should be represented uniquely. However, we find that this uniqueness cannot be modeled by bilinear based models, revealing the crack between the framework and models. To this end, we study the required conditions and propose a solution named Unit Ball Bilinear Model (UniBi). In addition to the theoretical superiority, UniBi is more robust and interpretable. Experiments demonstrate that UniBi models the uniqueness without the cost of performance and verify its a robustness and interpretability.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Submission Guidelines: Yes

Please Choose The Closest Area That Your Submission Falls Into: Deep Learning and representational learning