Common Neighbor Induced Message Passing for Inductive Link Prediction in Knowledge Graphs
Keywords: Inductive Link Prediction, Knowledge Graphs
Abstract: Inductive link prediction is a significant challenge in knowledge graphs, focusing on predicting potential relations between unseen entities during training. A promising approach is to utilize Graph Neural Networks (GNNs) to extract entity-independent features from surrounding subgraphs. However, existing mainstream subgraph extraction methods may lead to the loss of key entities and relations, resulting in many disconnected reasoning paths that seriously hinder effective message passing. To address this challenge, we propose a novel framework called Common Neighbor Induced Message Passing (CNMP), designed to enhance message passing even when reasoning paths are disconnected. We observe that the common neighbors of two entities must share a reasoning path. Based on this insight, CNMP enhances message passing by updating the distance labels of isolated common neighbors, even if they are unreachable. This allows CNMP to incorporate new connected equivalent relations, facilitating effective message passing. Furthermore, we introduce a CNMP+ strategy that further improves the preservation of entities and relations during the message-passing process. CNMP+ involves maintaining a list of common neighbors at various distances and using a probing strategy to reconstruct complete reasoning paths. Experiments across multiple datasets demonstrate that our method significantly outperforms existing state-of-the-art methods.
Primary Area: other topics in machine learning (i.e., none of the above)
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: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
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.
Submission Number: 7535
Loading