pycgs 1.1.0

Python implementation of the Coreference-based Graph Search (CGS) algorithm.

Coreference-based Graph Search (CGS)

This is the Python implementation of the CGS algorithm.

Documentation

The documentation for pycgs is available on the documentation website of the ShennongAlpha (ShennongDoc):

• English
• 中文

You can also contribute to the documentation on the ShennongDoc GitHub repository by submitting a pull request:

• English
• 中文

Foundational CGS

from pycgs import cgs

relationships = [('A', 'B'), ('B', 'C'), ('D', 'B'), ('E', 'F')]
primary_terms = cgs.foundational_cgs(relationships)

print(primary_terms)
# Output:
# {'A': 'C', 'B': 'C', 'C': 'C', 'D': 'C', 'E': 'F', 'F': 'F'}

Weighted CGS

from pycgs import cgs

weighted_relationships = [('A', 'B', 1), ('B', 'C', 2), ('D', 'B', 1), ('B', 'E', 1)]
primary_terms = cgs.weighted_cgs(weighted_relationships)

print(primary_terms)
# Output:
# {'A': 'C', 'B': 'C', 'C': 'C', 'D': 'C', 'E': 'E'}

PrimaryTermExtractor

PrimaryTermExtractor is a class that allows the extraction of primary terms from a given text based on a dictionary of coreference relationships between terms and their primary terms.

from pycgs.cgs import PrimaryTermExtractor

# Create a dictionary mapping terms to their primary terms
primary_term_dict = {
    "Artemisia annua Part-aerial": "nmm-0001",
    "Qing-hao": "nmm-0001",
    "黄花蒿地上部": "nmm-0001",
    "青蒿": "nmm-0001",
    "Ephedra sinica Stem-herbaceous": "nmm-0003",
    "Cao-ma-huang": "nmm-0003",
    "草麻黄草质茎": "nmm-0003",
    "草麻黄": "nmm-0003",
}

# Initialize the PrimaryTermExtractor
extractor = PrimaryTermExtractor(primary_term_dict)

# Extract primary terms from a mixed language text
text = "Both Artemisia annua Part-aerial and 草麻黄草质茎 are Natural Medicinal Materials and are used in traditional Chinese medicine."
result = extractor.extract_primary_terms(text)

print(result)
# Output:
# {'Artemisia annua Part-aerial': 'nmm-0001', '草麻黄草质茎': 'nmm-0003'}

Cite this work

@misc{yang2024shennongalphaaidrivensharingcollaboration,
      title={ShennongAlpha: an AI-driven sharing and collaboration platform for intelligent curation, acquisition, and translation of natural medicinal material knowledge}, 
      author={Zijie Yang and Yongjing Yin and Chaojun Kong and Tiange Chi and Wufan Tao and Yue Zhang and Tian Xu},
      year={2024},
      eprint={2401.00020},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2401.00020}, 
}