An LLM-Powered Semantic Alignment Framework for Journal Recommendation

2026-06-26

AI Summary

This study addresses the limited generalizability and interpretability of traditional journal recommendation methods, which often rely on supervised signals, handcrafted features, or historical interaction data. To overcome these limitations, the authors formulate journal recommendation as a semantic matching task between manuscript content and journal scope, proposing a training-free, large language model (LLM)-driven zero-shot recommendation framework. Built upon the DeepSeek-V3 LLM, the approach integrates article titles, abstracts, keywords, candidate journal descriptions, and reference information to infer suitability through a semantic alignment mechanism. Evaluated on a dataset of 23,609 statistics papers, the method achieves Top-3, Top-5, and Top-10 accuracy rates of 40.23%, 53.67%, and 70.05%, respectively, with a Jaccard similarity of 84% across repeated Top-5 recommendations, while also generating interpretable justifications that substantially enhance cross-scenario generalizability and stability.

Abstract

Journal recommendation is an important task in scholarly information systems. Existing approaches typically rely on supervised learning models, manually engineered features, or historical interaction data, which may limit their generalizability and interpretability. We propose an LLM-powered semantic alignment framework that formulates journal recommendation as a semantic matching problem between manuscript content and journal scope descriptions. The framework enables large language models (LLMs) to infer journal suitability directly from article titles, abstracts, keywords, and candidate journal information without task-specific training. Experiments are conducted using DeepSeek-V3 on a dataset of 23,609 articles from 49 journals in statistics and related fields. The proposed framework achieves Top-3, Top-5, and Top-10 accuracies of 40.23\%, 53.67\%, and 70.05\%, respectively. Additional analyses show that incorporating reference information generally improves recommendation performance and that recommendations remain highly stable across repeated runs, with an average Top-5 Jaccard similarity of 84\%. The framework also generates interpretable reasoning outputs that provide insights into the recommendation process. These findings demonstrate the potential of LLMs as a training-free and scalable paradigm for journal recommendation and scholarly decision support.

Problem

• journal recommendation
• semantic alignment
• large language models
• scholarly information systems
• generalizability

Innovation

• LLM-powered
• semantic alignment
• journal recommendation
• training-free
• interpretable reasoning