Bodhi

As of June 17, 2024, Bodhi is a component of Scholarly

The goal of Bodhi is to empower novice researchers to quickly understand a field of research. With Bodhi, users can discover the most pivotal papers in a specific area and read them in order of complexity.

How it Works

The process consists of four main steps. Because I thought of them in the shower, they are subject to change:

1. Embed Research Papers:

   • Embed the metadata of research papers and perform clustering to find different research areas.

2. Filter Important Papers:

   • Within each cluster, filter for the top n most important papers. The importance score reflects how necessary it is to read a particular paper to understand the research in its cluster.
   • For example, a seminal paper would have a high importance score.

3. Create Directed Graphs:

   • Create a directed graph for each sparsified cluster, ensuring that edges flow from simpler papers to more complex ones.

4. Find Minimum Spanning Arborescence (MSA):

   • Find the MSA for each directed graph using the Chu-Liu/Edmonds/Bock Algorithm. The resulting arborescence serves as a reading path for researchers.
   • For a detailed explanation of the Chu-Liu/Edmonds/Bock Algorithm, refer to Section 2.2.

Integration with Scholarly

Bodhi is a component of Scholarly. Scholarly simplifies the literature review process for researchers by providing in-line explanations, abstract-to-text hyperlinks, and a recommendation engine for important current and future research papers.

More coming soon :)

Usage

To use this tool, follow the steps below:

1. Clone the repository:

   git clone https://github.com/RGBmarya/scholarly.git

2. Navigate to the project directory:

   cd scholarly

3. Follow the instructions in the Jupyter Notebook (main.ipynb) to embed research papers, filter important papers, create directed graphs, and find the MSA for each cluster.

Contribution

Feel free to contribute to the project by submitting a pull request or opening an issue.

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Thanks for trying Bodhi! (and shoutout to those who get the reference). Much ❤️.

- Mihir