An Interactive and Personalized literature survey generation system.
| 2025/7/8 | 🚀 Our InteractiveSurvey demo is now officially online! Experience the power of AI-driven survey paper generation with our interactive web platform. Try it now! 🌟 |
| 2025/4/15 | The paper InteractiveSurvey: An LLM-based Personalized and Interactive Survey Paper Generation System is now released! 🎉✨ |
InteractiveSurvey is an interactive and personalized tool designed to help researchers efficiently conduct literature reviews. By leveraging natural language processing (NLP) and Large Language Models (LLMs), it enables users to collect, organize, and generate structured literature surveys effortlessly.
- 📝 Automatic Literature Review Generation: Extract key insights from papers and generate structured literature surveys.
- 💡 Interactive Exploration: Dynamically filter, refine, and customize your survey in real-time.
- 📄 PDF Export: Easily generate high-quality literature surveys in PDF format with either Markdown or LaTeX.
- ⚡ Multimodality: Extract figures from references and insert customized figures by yourself.
- 🐳 Docker Support: Quickly deploy and run the application in a containerized environment.
demo_2.mp4
📺 Demo Video: You can watch the demo video at https://www.bilibili.com/video/BV1dju2z8ESw/?spm_id_from=333.1387.homepage.video_card.click
example_md.mp4
example_latex.mp4
Interactive requires Python 3.10. A minimum 20G disk space is required
Clone the repository to your local machine:
git clone https://github.com/TechnicolorGUO/InteractiveSurvey
cd InteractiveSurveyCreate a virtual environment and activate it:
conda create -n interactivesurvey python=3.10
conda activate interactivesurveyInstall the required dependencies:
python scripts/setup_env.pyFor the ConnectTimeout error when downloading Huggingface models, please run the following script:
pip install modelscope
wget https://gcore.jsdelivr.net/gh/opendatalab/MinerU@master/scripts/download_models.py -O download_models.py
python download_models.pyCreate a .env file in the root directory of the project and add the following configurations:
OPENAI_API_KEY=<your_openai_api_key_here>
OPENAI_API_BASE=<your_openai_api_base_here>
MODEL=<your_preferred_model_here>Replace the placeholders with your actual OpenAI API key, API base URL, and preferred model.
To test the system on GPU, you also need to follow the instructions provided by MinerU:
Start the development server by running the following command:
python src/manage.py runserver 0.0.0.0:8001(Replace 8001 with any available port number of your choice.)
Once the server is running, open your browser and navigate to:
http://localhost:8001
You can now use the Auto Literature Survey Generator to upload, analyze, and generate literature surveys!
Before proceeding, ensure you have cloned the repository and configured your .env file in the root directory of the project. The .env file must include the following configurations:
OPENAI_API_KEY=<your_openai_api_key_here>
OPENAI_API_BASE=<your_openai_api_base_here>
MODEL=<your_preferred_model_here>Replace the placeholders with your actual OpenAI API key, API base URL, and preferred model.
If you have GPU support, you can build and run the GPU version of the Docker container using the following commands:
# Build the Docker image
docker build -t interactivesurvey .
# Run the Docker container (with GPU support)
docker run --gpus all -p 8001:8001 interactivesurveyIf you do not have GPU support, you can run the CPU version of the Docker container. Note: Before building and running, you need to manually remove the following line from the scripts/additional_scripts.py file:
"device-mode": "cuda",Then run the following commands:
# Build the Docker image
docker build -t interactivesurvey-cpu .
# Run the Docker container (with CPU support)
docker run -p 8001:8001 interactivesurvey-cpuAfter starting the container, access (http://localhost:8001)[http://localhost:8001] to confirm that the application is running correctly.
If you want to generate surveys directly without using the frontend, follow these steps:
- Navigate to the
src/demo/survey_generation_pipelinedirectory:
cd src/demo/survey_generation_pipeline- Copy the
.envfile to this directory. If you already have a .env file in the root of your project, you can copy it like this:
cp ../../../.env .Note: Ensure the .env file contains the required configurations (e.g., OPENAI_API_KEY, OPENAI_API_BASE, and MODEL).
- Run the pipeline directly:
python main.pyThis will execute the survey generation pipeline on our sample PDFs and output the results (.md and .pdf) to the result folder directly.
- Modify the script for your own sample
The
main.pycontains the following code to generate a survey:
if __name__ == "__main__":
root_path = "."
pdf_path = "./sample_pdfs" #Set this to the path of the folder containing your PDF files.
survey_title = "Automating Literature Review Generation with LLM" #Set this to the title of your survey.
cluster_standard = "method" #Set this to the clustering standard you want to use.
asg_system = ASG_system(root_path, 'test', pdf_path, survey_title, cluster_standard) #test refers to the survey_id which prevent you from parsing pdfs again.
asg_system.download_pdf()
# Downloads PDFs to "./sample_pdfs". Add with your own files for upload.
asg_system.parsing_pdfs()
asg_system.description_generation()
asg_system.agglomerative_clustering()
asg_system.outline_generation()
asg_system.section_generation()
asg_system.citation_generation()If you have any enquiries, please email guobeichen0228@gmail.com