A tool to map FrameNet Lexical Units to BabelNet synsets using the distance between sentence embeddings of corresponding definitions
Create a virtual environment and install dependencies:
conda env -n mapper python=3.7
conda activate mapper
cd FnSenseMapper
pip install -r requirements.txt
Run the mapper:
python run.py
Note that the existing bn_edges.csv
in resources
is truncated due to Github's file size restrictions.
The following mapper parameters may be modified in config.yml
:
- Maximum number of candidates to keep
- Cosine similarity cutoff threshold
- Whether to discard candidates where LU lemma is not a (partial) substring of BN entry name