Taxoble

Workflow

1. Optional: Download PubMed Data

Use download_data_from_pubmed.ipynb to fetch PubMed records.

You can skip this step if you already have a CSV file with candidate terms.

2. Preprocess Candidate Terms

Edit the configuration block at the bottom of preprocess_input.py.

Then run:

python preprocess_input.py

This creates keyword_mapping.pkl.

Preprocessing modes:

• canonicalize: expands abbreviations and merges synonyms.
• canonicalize_and_filter_by_domain: also removes terms that are not relevant to the configured domain.

3. Construct the Taxonomy

Open construct_taxonomy.ipynb and edit the TaxonomyConfig cell.

Run the notebook cells in order.

Outputs

Final outputs are written to:

results/<folder_name>/<experiment_name>/

The directory contains:

• final_taxonomy.txt: the readable taxonomy.
• edges.json: the taxonomy edge list, represented as (child, parent) pairs.
• nodes.json: the complete list of taxonomy nodes.
• term_to_node_mappings.json: the mapping from terms to their assigned taxonomy nodes.

Notes

• The pipeline relies on LLM. Review final outputs before using them in downstream analyses.
• Some cycle resolution may require manual input when the LLM cannot confidently repair a graph.
• Large keyword sets can require substantial memory for embedding and similarity computations.

Citation

If you use this code in a publication or shared project, please cite the repository.