https://github.com/morningmoni/TaxoRL

Code for paper "End-to-End Reinforcement Learning for Automatic Taxonomy Induction", ACL 2018
https://github.com/morningmoni/TaxoRL

dynet reinforcement-learning taxonomy taxonomy-construction taxonomy-induction wordnet

Last synced: about 1 month ago
JSON representation

Code for paper "End-to-End Reinforcement Learning for Automatic Taxonomy Induction", ACL 2018

• Host: GitHub
• URL: https://github.com/morningmoni/TaxoRL
• Owner: morningmoni
• License: mit
• Created: 2018-05-08T16:04:53.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-10-10T15:36:06.000Z (about 6 years ago)
• Last Synced: 2024-08-05T03:01:50.713Z (5 months ago)
• Topics: dynet, reinforcement-learning, taxonomy, taxonomy-construction, taxonomy-induction, wordnet
• Language: Python
• Homepage:
• Size: 2.94 MB
• Stars: 65
• Watchers: 6
• Forks: 13
• Open Issues: 2
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-taxonomy - https://github.com/morningmoni/TaxoRL

README

        
# TaxoRL

code for "End-to-End Reinforcement Learning for Automatic Taxonomy Induction" ACL 2018 [[arXiv]](https://arxiv.org/abs/1805.04044)

## Requirements

python **2.7**  

dynet 2.0  

tqdm  

## Data

**Preprocessed pickled data including everything else for the WordNet data can be downloaded [here](https://drive.google.com/file/d/1EXeMb69fcoQgiNORAXcg2vZPR7yBbjrY/view?usp=sharing).**

Preprocessed pickled data including everything else for the WordNet data and SemEval-2016 can be downloaded [here](https://drive.google.com/file/d/1p70QAe9yYD1kEAeDyjPvnfJKILaS2ZRl/view?usp=sharing). If you run on SemEval-2016, use [dev_twodatasets.tsv](https://drive.google.com/file/d/1n3XuwiXe3HQAl3ogDV0VI3FNe5MwOYt4/view?usp=sharing) instead of dev_wnbo_hyper.tsv.  Caution: it may take 40+ GB memory.

Go to https://morningmoni.github.io/wordnet-vis/ to see the visualization of WordNet subtrees.

**DIY** 

- To build everything from scratch, first download corpora such as Wikipedia, [UMBC](https://ebiquity.umbc.edu/resource/html/id/351/UMBC-webbase-corpus), and [1 Billion Word Language Model Benchmark](http://www.statmt.org/lm-benchmark/).

- To preprocess the corpus, generate a vocabulary file and use the scripts under ./corpus/ to find dependency paths between terms in the vocabulary. The scripts are modified based on [LexNET](https://github.com/vered1986/LexNET). Instructions can be found [here](https://github.com/vered1986/LexNET/wiki/Detailed-Guide). It may take several hours to finish this process. 

- Run train_RL.py and it will compute all the features and save them into pickle files.

## Run

Run train_RL.py for training and testing. All the parameters are in *argparse* and have default values so that you can run without specifying any parameters (but feel free to tune them).

In each epoch, the performance on training/validation/test sets is reported. You may exit the program at any time.

## Cite


@InProceedings{P18-1229,

  author = 	"Mao, Yuning

		and Ren, Xiang

		and Shen, Jiaming

		and Gu, Xiaotao

		and Han, Jiawei",

  title = 	"End-to-End Reinforcement Learning for Automatic Taxonomy Induction",

  booktitle = 	"Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",

  year = 	"2018",

  publisher = 	"Association for Computational Linguistics",

  pages = 	"2462--2472",

  location = 	"Melbourne, Australia",

  url = 	"http://aclweb.org/anthology/P18-1229"

}