Journal Recommendation Agent

A prompt-based journal recommendation system powered by LLMs. Given a manuscript's title, abstract, and keywords, the agent recommends the top-10 most suitable journals from a predefined candidate list based on semantic compatibility.

---

Project Structure

journal-agent/
├── agent.py                 # Core Agent class
├── prompt.py                # Prompt construction module
├── example.py               # Interactive usage script
├── data/
│   ├── journal_scope.csv    # Journal scope descriptions
│   └── papers.csv           # (Optional) Input papers for batch recommendation
├── requirements.txt         # Dependencies
└── README.md

---

Requirements

• Python 3.8+
• An API key from DeepSeek or OpenAI

Install dependencies:

pip install -r requirements.txt

---

Data Format

journal_scope.csv

Must contain exactly two columns:

Journal	Scope
Annals of Statistics	Publishes research papers in all areas of statistics...
Journal of Machine Learning Research	Covers all areas of machine learning...

papers.csv (for batch recommendation)

Must contain the following three columns:

title	abstract	keywords
A Bayesian Approach to...	This paper proposes...	Bayesian, inference, statistics

---

Usage

Step 1: Configure your API key

Open example.py and fill in your API key and file paths:

API_KEY = "your-api-key-here"
SCOPE_CSV_PATH = "data/journal_scope.csv"
BASE_URL = "https://api.deepseek.com/v1"  # or https://api.openai.com/v1 for OpenAI

Step 2: Run the agent

python example.py

You will be prompted to choose a mode:

==================================================
   Journal Recommendation Agent
==================================================
Please select a mode:
  [1] Single article recommendation
  [2] Batch recommendation (from CSV)
==================================================

---

Example Output

Single Article Mode

Top-10 Recommended Journals:
----------------------------------------
  RECOMMENDATION_1:  Annals of Statistics
  RECOMMENDATION_2:  Journal of the American Statistical Association
  RECOMMENDATION_3:  Biometrika
  RECOMMENDATION_4:  Journal of the Royal Statistical Society
  RECOMMENDATION_5:  Statistica Sinica
  ...

Batch Mode

Loaded 100 articles. Starting batch recommendation...
Processing Journal Recommendations: 100%|████████| 100/100
Done! Results saved to: data/results.csv

The output CSV will contain the original columns plus 10 additional columns: RECOMMENDATION_1, RECOMMENDATION_2, ..., RECOMMENDATION_10.

---

Using the Agent Programmatically

You can also import and use the agent directly in your own script:

from agent import JournalRecommendAgent

agent = JournalRecommendAgent(
    api_key="your-api-key-here",
    scope_csv_path="data/journal_scope.csv"
)

# Single recommendation
result = agent.recommend(
    title="Your paper title",
    abstract="Your abstract here...",
    keywords="keyword1, keyword2, keyword3"
)
for k, v in result.items():
    print(f"{k}: {v}")

# Batch recommendation
import pandas as pd
df = pd.read_csv("data/papers.csv")
df_result = agent.recommend_batch(df)
df_result.to_csv("data/results.csv", index=False)

---

Notes

• The agent sets temperature=0 by default to ensure stable and reproducible outputs.
• Batch mode uses multithreading (max_workers=5 by default) to speed up large-scale inference.
• The recommendation is based solely on the provided manuscript information. No external knowledge, web search, or prior publication history is used.
• The candidate journal list is fixed and defined by journal_scope.csv.

---

Citation

If you use this tool in your research, please cite:

@article{paper2026,
  title   = {Paper Title},
  author  = {Name},
  journal = {Journal},
  year    = {2026}
}

---

License

This project is licensed under the MIT License.

Maintenance

The candidate journal list (data/journal_scope.csv) will be updated periodically to reflect changes in journal scopes and expansions of the journal pool. We recommend checking the latest release for the most up-to-date version.