Commonsense Causal Reasoning between Short Texts

1. CausalNet

CausalNet consists of a large amount of causal relationships extracted from Bing web pages.

Each causal relationship is a triple as following: CAUSE_WORD[\t]EFFECT_WORD[\t]FREQUENCY

You can download CausalNet from https://adapt.seiee.sjtu.edu.cn/causal/

If you have any questions, feel free to contact Zhiyi Luo at jessherlock@sjtu.edu.cn . – Zhiyi Luo, Feb 24th, 2017.

2. Publications

Please cite the following paper if you are using CausalNet and the code. Thanks!

• Zhiyi Luo, Yuchen Sha, Kenny Q. Zhu, Seung-won Hwang, Zhongyuan Wang, “Commonsense Causal Reasoning between Short Texts”, Proc. of 15th Int. Conf. on Principles of Knowledge Representation and Reasonging (KR’2016), Cape Town, South Africa.

3. Quick Start

This repository is an implementation of the approach proposed in “Commonsense Causal Reasoning between Short Texts”, KR’2016.

Follow these steps to get started:

1. Download CausalNet from https://adapt.seiee.sjtu.edu.cn/causal/ ,then tar -xjf cs.tar.bz2.

2. Download the KR-COPA.jar from https://adapt.seiee.sjtu.edu.cn/causal/tools/KR-COPA.jar .

3. Create the Log folder: mkdir -p Log

4. Set YOUR PATH in light-copa-config.ini

5. Run java -Xmx25g -cp KR-COPA.jar edu.sjtu.copa.exe.COPAEvaluation