Communicative Message Passing for Inductive Relation Reasoning
نویسندگان
چکیده
Relation prediction for knowledge graphs aims at predicting missing relationships between entities. Despite the importance of inductive relation prediction, most previous works are limited to a transductive setting and cannot process previously unseen The recent proposed subgraph-based reasoning models provided alternatives predict links from subgraph structure surrounding candidate triplet inductively. However, we observe that these methods often neglect directed nature extracted weaken role information in modeling. As result, they fail effectively handle asymmetric/anti-symmetric triplets produce insufficient embeddings target triplets. To this end, introduce Communicative Message Passing neural network Inductive reLation rEasoning, CoMPILE, reasons over local structures has vigorous bias entity-independent semantic relations. In contrast existing models, CoMPILE strengthens message interactions edges entitles through communicative kernel enables sufficient flow information. Moreover, demonstrate can naturally relations without need explosively increasing number model parameters by extracting enclosing subgraphs. Extensive experiments show substantial performance gains comparison state-of-the-art on commonly used benchmark datasets with variant settings.
منابع مشابه
Modular Reasoning for Message-Passing Programs
Verification of concurrent systems is difficult because of the inherent nondeterminism. Modern verification requires better locality and modularity. Reasoning of shared memory systems has gained much progress in these aspects. However, modular verification of distributed systems is still in demand. In this paper, we propose a new reasoning system for message-passing programs. It is a novel logi...
متن کاملRely-Guarantee Based Reasoning for Message-Passing Programs1
The difficulties of verifying concurrent programs lie in their inherent non-determinism and interferences. Rely-Guarantee reasoning is one useful approach to solve this problem for its capability in formally specifying interthread interferences. However, modern verification requires better locality and modularity. It is still a great challenge to verify a message-passing program in a modular an...
متن کاملRely-Guarantee Based Reasoning for Message-Passing Programs
The difficulties of verifying concurrent programs lie in their inherent non-determinism and interferences. Rely-Guarantee reasoning is one useful approach to solve this problem for its capability in formally specifying inter-thread interferences. However, modern verification requires better locality and modularity. It is still a great challenge to verify a message-passing program in a modular a...
متن کاملReasoning MPE to Multiply Connected Belief Networks Using Message Passing
Finding the I Most Probable IJxplanations (MPE) of a given evidence, Se, in a Bayesian belief network is a process to identify and order a set of composite hypotheses, His, of which the posterior probabilities are the I largest; i.e., Pr(Hii&) 2 Pr(H21&) 2 . . . L Pr(&ISlz)~ A composite hypothesis is defined as an instantiation of all the non-evidence variables in the network. It could be shown...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Proceedings of the ... AAAI Conference on Artificial Intelligence
سال: 2021
ISSN: ['2159-5399', '2374-3468']
DOI: https://doi.org/10.1609/aaai.v35i5.16554