Retrieval-guided Twin Fusion with Similarity-aware Contrast for Molecule-Text Alignment

2026-08-16

AI Summary

This study addresses the lack of fine-grained semantic alignment between molecular substructures and textual descriptions in molecule-text matching. To overcome this limitation, we propose RISEN, a novel framework that innovatively integrates a retrieval-guided twin fusion mechanism with similarity-aware soft-threshold contrastive learning. By constructing latent twin molecules to augment substructure semantic representations, RISEN achieves precise cross-modal fine-grained alignment. Extensive experiments demonstrate that RISEN significantly outperforms existing baselines across multiple benchmark datasets, effectively enhancing downstream tasks such as molecular search and property prediction. Consequently, this work establishes a new paradigm for cross-modal understanding in molecular science, bridging the gap between structural and textual modalities through enhanced semantic correspondence.

Abstract

This paper studies the problem of molecule-text alignment, which aims to project molecules and their textual descriptions into a joint latent space for downstream tasks including molecule search and molecular property prediction. Previous approaches typically combine graph structure mining with contrastive learning to enhance joint representation learning. However, they typically neglect fine-grained semantic relationships between substructures and texts, leading to suboptimal performance on downstream tasks. Towards this end, we propose a novel approach named Retrieval-guided Twin Fusion with Similarity-aware Contrast (RISEN) for molecule-text alignment. The core idea of RISEN is to construct a latent twin molecule for each substructure with cross-modal retrieval for semantic enhancement. In particular, for each substructure query, we retrieve relevant textual descriptions and sample several molecules that share similar descriptions of substructures. Then, we aggregate their representations via attention pooling for a twin latent representation, which would be further fused with the original substructure for representation enrichment. In addition, we measure the similarity across substructures and texts, which would further guide cross-modal contrastive learning with soft thresholding. Extensive experiments on benchmark datasets validate the superiority of the proposed RISEN in comparison with existing baselines.

Problem

• Molecule-Text Alignment
• Fine-grained Semantic Relationships
• Substructure-Text Matching
• Joint Representation Learning

Innovation

• Retrieval-guided Twin Fusion
• Similarity-aware Contrast
• Molecule-Text Alignment
• Cross-modal Retrieval
• Soft Thresholding