re-RL-Crowd

Introduction

It is based on RE-RL model. The following have been modified:

1. It dosen't use entity embeddings as the input features
2. Construct the bag of sentences randomly split sentences like batch data
3. Set the sampling size(number of pre-actions at training the policy network) to 1
4. Use twitter tokenizer to adjust Korean training data

Licenses

• CC BY-NC-SA Attribution-NonCommercial-ShareAlike
• If you want to commercialize this resource, please contact to us

Publisher

Machine Reading Lab @ KAIST

Acknowledgement

This work was supported by Institute for Information & communications Technology Promotion(IITP) grant funded by the Korea government(MSIT) (2013-0-00109, WiseKB: Big data based self-evolving knowledge base and reasoning platform)