NGTM: Substructure-based Neural Graph Topic Model for Interpretable Graph Generation

TL;DR

NGTM introduces a graph generation framework that models graphs as mixtures of interpretable latent topics, enhancing transparency and control over the generated structures.

Contribution

The paper presents NGTM, a novel neural model that incorporates topic modeling principles into graph generation for improved interpretability and controllability.

Findings

Achieves competitive graph generation quality.

Enables fine-grained control over structural features.

Provides explicit interpretability at local and global levels.

Abstract

Graph generation plays a pivotal role across numerous domains, including molecular design and knowledge graph construction. Although existing methods achieve considerable success in generating realistic graphs, their interpretability remains limited, often obscuring the rationale behind structural decisions. To address this challenge, we propose the Neural Graph Topic Model (NGTM), a novel generative framework inspired by topic modeling in natural language processing. NGTM represents graphs as mixtures of latent topics, each defining a distribution over semantically meaningful substructures, which facilitates explicit interpretability at both local and global scales. The generation process transparently integrates these topic distributions with a global structural variable, enabling clear semantic tracing of each generated graph. Experiments demonstrate that NGTM achieves competitive generation quality while uniquely enabling fine-grained control and interpretability, allowing users to tune structural features or induce biological properties through topic-level adjustments.