searching for software promises in grant applications
Project description
soft-search
searching for software promises in grant applications
Installation
Stable Release: pip install soft-search
Development Head: pip install git+https://github.com/si2-urssi/eager.git
Quickstart
- Load our best model (the "TF-IDF Vectorizer Logistic Regression Model")
- Pull award abstract texts from the NSF API
- Predict if the award will produce software using the abstract text for each award
from soft_search.constants import NSFFields, NSFPrograms
from soft_search.label import load_soft_search_model
from soft_search.nsf import get_nsf_dataset
# Load the model
pipeline = load_soft_search_model()
# Pull data
data = get_nsf_dataset(
start_date="2022-05-01",
end_date="2022-07-01",
program_name=NSFPrograms.Computer_and_Information_Science_and_Engineering,
dataset_fields=[NSFFields.id_, NSFFields.abstractText],
require_project_outcomes_doc=False,
)
# Predict
data["prediction"] = pipeline.predict(data[NSFFields.abstractText])
print(data)
# abstractText id prediction
# 0 Human AI Teaming (HAT) is an emerging and rapi... 2213827 software-not-predicted
# 1 This project furthers progress in our understa... 2213756 software-predicted
Annotated Training Data
from soft_search.data import load_soft_search_2022
df = load_soft_search_2022()
Reproducible Models
To train and evaluate all of our models you can run the following:
pip install soft-search
fit-and-eval-all-models
Also available directly in Python
from soft_search.label.model_selection import fit_and_eval_all_models
results = fit_and_eval_all_models()
Documentation
For full package documentation please visit si2-urssi.github.io/eager.
Development
See CONTRIBUTING.md for information related to developing the code.
MIT License
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
soft-search-0.5.0.tar.gz
(953.6 kB
view hashes)
Built Distribution
soft_search-0.5.0-py3-none-any.whl
(957.6 kB
view hashes)
Close
Hashes for soft_search-0.5.0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 76ebd6ab042ea073092f2744dbac43c39160f1526dce0be0e1e5c66794374582 |
|
MD5 | 2a90e6649604405dbaecd0e2ad3c6ea9 |
|
BLAKE2b-256 | 6a012738159053c733478acb54ea5003d97c63fbdb1f59e9db1a6ae60c99eed9 |