helloworld0909 / event-chains-extractor

Implementation of event chains extraction algorithm using Dependency Parsing & Coreference Resolution of CoreNLP

event-chains-extractor

Implementation of event chains extraction algorithm using Dependency Parsing & Coreference Resolution via CoreNLP

Input

Gigaword English NY Times (SGML format version)

Build with Maven

Compile

mvn compile

or

Build jar

mvn package

or

Build jar with dependencies

mvn package assembly:single

Run

1. Adjust input path and put input corpus files like this:

   data/nyt_eng_199407

2. Convert SGML format to JSON

   cd scripts
   python sgml2json.py
   

3. Run extraction program

   java -jar finalpj-0.1-XXXXX.jar
   

   or

   mvn exec:java -Dexec.mainClass="com.helloworld09.nlp.EventChain"
   

Reference

1. Chambers N, Jurafsky D. Unsupervised learning of narrative event chains[C]//ACL. 2008, 94305: 789-797.
2. Wang Z, Zhang Y, Chang C Y. Integrating Order Information and Event Relation for Script Event Prediction[C]//Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing. 2017: 57-67.
3. Granroth-Wilding M, Clark S. What Happens Next? Event Prediction Using a Compositional Neural Network Model[C]//AAAI. 2016: 2727-2733.