Bridging the Gap Between Molecule and Textual Descriptions via Substructure-aware Alignment

2025-10-30

AI Summary

Existing molecular-text representation learning methods struggle to model fine-grained alignment between molecular substructures and chemical phrases, limiting cross-modal semantic matching accuracy. To address this, we propose MolBridge—a novel framework that enables precise substructure-to-phrase alignment for the first time. MolBridge jointly optimizes graph neural networks and language models via substructure-aware contrastive learning. It further introduces phrase-level alignment enhancement and a self-refinement mechanism to dynamically filter noisy alignment signals. Evaluated on multiple molecular-text retrieval and cross-modal understanding benchmarks, MolBridge consistently outperforms state-of-the-art methods by significant margins. Ablation studies confirm that fine-grained substructure–phrase alignment is critical for enhancing multimodal representation capability. Our work establishes a new paradigm for interpretable and semantically grounded molecular-text modeling.

Technology Category

• Natural Language Processing: Language Grounding & Multi-modal NLP
• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Molecule and text representation learning has gained increasing interest due to its potential for enhancing the understanding of chemical information. However, existing models often struggle to capture subtle differences between molecules and their descriptions, as they lack the ability to learn fine-grained alignments between molecular substructures and chemical phrases. To address this limitation, we introduce MolBridge, a novel molecule-text learning framework based on substructure-aware alignments. Specifically, we augment the original molecule-description pairs with additional alignment signals derived from molecular substructures and chemical phrases. To effectively learn from these enriched alignments, MolBridge employs substructure-aware contrastive learning, coupled with a self-refinement mechanism that filters out noisy alignment signals. Experimental results show that MolBridge effectively captures fine-grained correspondences and outperforms state-of-the-art baselines on a wide range of molecular benchmarks, highlighting the significance of substructure-aware alignment in molecule-text learning.

Problem

• Aligns molecular substructures with chemical phrases
• Captures fine-grained molecule-text correspondences
• Improves molecular representation learning via substructure alignment

Innovation

• Substructure-aware alignment for molecule-text learning
• Substructure-aware contrastive learning with self-refinement
• Augmenting molecule-description pairs with alignment signals