coleridge-rich-context-larc

Code Repository for RCC submission

Dependencies

Scala for AllenAI Science Parse Python dependencies:

• python 3.6 via Anaconda
• scikit-learn 0.20.1
• pandas

Project Structure

This project contains few important folders:

• models: contains all learning models.
• resources: contains resource files such as SAGE research methods and research fields.
• tools: contains external packages such as AllenAI Science Parse.
• data: contains dataset (for training only).

All main codes are under project/ folder.

AllenAI Science Parse tool

AllenAI Science Parse parses PDF publication papers into following fields: title, authors, abstract, sections, bibliography.

How to run

java -Xmx6g -jar tools/science-parse-cli-assembly-2.0.2-SNAPSHOT.jar -o ../data/input/files/json/ ../data/input/files/pdf/

CLI Snippets

Datasets prediction

python datasets_predict.py --input_dir ../data/input/ --output_dir ../data/output/

Research methods recommendation

python rmethods_rec.py --input_dir ../data/input/ --output_dir ../data/output/

Research fields recommendation

python rfields_rec.py --input_dir ../data/input/ --output_dir ../data/output/

Following snippets are examples for training models

Train dataset detection

python dataset_detect_train.py --input_dir data/train_test/  --output models/dataset_detect.model

Train research methods recommendation models

python rmethods_rec_train.py --input data/rmethod_ctx_train_mx100.json --output models/rmethods_rec.model

Train research fields recommendation models

python rfields_rec_train.py --input data/rfield_10.json --output models/rfields_rec.model