SUAEx

The "Simple Unsupervised Similarity-based Aspect Extraction" is a method to extract aspect words(most relevant words in a given sentence and domain). SUAEx implements a simple approach that only relies on word-embeddings similarity with respect to reference words. Furthermore, it emulates the attention mechanism of neural networks by using only the similarity of words.

Dependencies

* python3 
* gensim 3.5.2
* sklearn 0.0

Usage

Model

* Download the word-embeddings model(restaurant.txt) from: 
		https://mega.nz/#!rihQiYhL!jdGipAwlxX4F-RWTRjoNQLZWH_fit2zwQZBCn8QsQxc
* Create the folder "models"
* Put the download model on the created folder

This folder contains the implementation of SUAEx which is organized in three folders

* word_simils
* category_atribution
* select_aspects

Through this implementation, we can run SUAEx on the restaurant ABAE dataset. Bellow the steps to run the example

* On "word_simils/code/" run the python script  getaspectbysimil_v3restaurant.py
		*python getaspectbysimil_v3restaurant.py*
* On "category_atribution/" run the python script cat_atrib_rest.py
		*python cat_atrib_rest.py*
* On "select_aspects/" run the python script select.py
		*python select.py*		

Citation

If using SUAEx, please cite our work by : 
	@inproceedings{Suarez19, 
  		title={Simple Unsupervised Similarity-Based Aspect Extraction}, 
  		author={Danny Suarez Vargas, Lucas R. C. Pessutto, and Viviane Pereira Moreira}, 
  		booktitle={20th International Conference on Computational Linguistics and Intelligent Text Processing (CICLing)}, 
  		year={2019} 
	}