InductWave: Inductive Multi-Hop Logical Query Answering on Knowledge Graphs

2026-07-08

AI Summary

Existing approaches to multi-hop logical query answering over knowledge graphs are predominantly transductive, limiting their ability to generalize to unseen entities and hindering scalability in large-scale settings. This work proposes InductWave, the first method to integrate wavelet transforms into inductive logical query answering. By leveraging wavelet basis embeddings within an inductive graph neural network architecture, InductWave supports complex Existential First-Order (EFO) queries involving conjunction, disjunction, and negation. The approach substantially reduces the required number of message-passing layers while consistently outperforming state-of-the-art models across varying training–testing splits on FB15k-237, achieving superior performance with only 75% of the typical layer depth. Furthermore, InductWave successfully scales to ultra-large knowledge graphs such as Wiki-KG, demonstrating its practical applicability and efficiency.

Abstract

Logical Multi-Hop Query Answering over Knowledge Graphs (KGs) can be formulated as querying, with an implicit completeness assumption. Current works mainly focus on Existential First Order Logic (EFO) queries. These EFO queries contain conjunction, disjunction, and negation operators. Most existing works employ transductive reasoning, meaning they are not capable of reasoning over entities unseen during training. In the real world, there is a resource scarcity, and we cannot train a model with all the nodes of a large KG. Hence, we propose InductWave, a wavelet-based inductive embedding method for logical query answering on large KGs. Here, the training graph consists of fewer nodes than the test graph. Our model performs on par with the baseline models while having half the number of message-passing layers. It outperforms all of them in most cases, with 75% of the layers. These fewer resource requirements enable us to evaluate InductWave on massive graphs, such as Wiki-KG. We test our model using extensive experiments across varying train-test graph proportions of the FB15k-(237) dataset, comparing it with the state-of-the-art models. The code and datasets for the model are available at https://github.com/kracr/inductwave/.

Problem

• inductive reasoning
• multi-hop query answering
• knowledge graphs
• logical query
• unseen entities

Innovation

• inductive reasoning
• wavelet-based embedding
• multi-hop logical query answering
• knowledge graphs
• message-passing efficiency