🔋🔍 NLP4SIB: Extracting a Database of Challenges and Mitigation Strategies for Sodium-ion Battery Development

---

🌟 Key Highlights

• Database of Challenges: A detailed compilation of the challenges faced in the performance and synthesis of SIB CAMs: database/challenge_database.xlsx
• Mitigation Strategies: Identification and pairing of challenges with potential mitigation strategies: database/mitigation_database.xlsx

🖥️ NLP Pipeline

The sequential filtering and visualization pipeline utilizing sentence classification, phrase-level classification, and relationship extraction.

📩 Contact

For inquiries or further information, please contact: [email protected]

🙌 Acknowledgements

Special thanks to Vineeth Venugopal, Elsa Olivetti, Kevin J. Huang, Ryan Stephens, and MIT for their support.

---

🤖 Train the Sentence Classifier

Step 1: Create a new Env

Create a new environment & activate** named nlp4sib using Python 3.8:

conda create -n nlp4sib-sentence python==3.8
conda activate nlp4sib-sentence

Step 2: Environment Setup

Install the necessary dependencies** from the requirements.txt file:

pip install -r requirements.txt

Step 3: Start Training

Train the Sentence Classifier**:

python classifiers/sentence_classifier/train_bert.py 

🔍 Running Entity & Relation Extraction with DYGIE++

Step 1: Initialize DyGIE++

First, initialize the DyGIE++ submodule with the following commands:

git submodule init
git submodule update

Step 2: Environment Setup

Create and activate a new Conda environment for the project:

conda create -n nlp_4sib-phrase python==3.8
conda activate nlp_4sib-phrase

Navigate to the DyGIE++ directory and install the required dependencies:

cd classifiers/phrase_classifier/dygiepp
pip install -r requirements.txt
pip install transformers==4.2.1 # being able to use recent models
pip install numpy==1.19.0 # version conflict
conda develop .
cd ../../..

Step 3: Training DyGIE++

Train the DyGIE++ model using the labeled sentences. Two training configurations are provided below:

• Improvement Configuration:

  allennlp train "classifiers/phrase_classifier/best_parameter_study_improvement.json" \
      --serialization-dir "classifiers/phrase_classifier/dygiepp/models/improvement" \
      --include-package dygie

• Challenge Configuration:

  allennlp train "classifiers/phrase_classifier/best_parameter_study_challenge.json" \
      --serialization-dir "classifiers/phrase_classifier/dygiepp/models/challenge" \
      --include-package dygie