Automatically assign semantics to large data sets from heterogeneous sources based on their features using several Statistical and Machine Learning techniques.

1. Elasticsearch
2. Pyspark
3. scikit-learn
4. pandas

1. Build docker image

cd container; docker build -t isi/semantic-labeling .

2. Start elasticsearch:

docker-compose up

3. Calling API

bin/semantic_labeling.sh <train_dataset> <test_dataset> <train_dataset2>