phiradet / aspect_term_extraction

Aspect Term Detector

This repository contains Jupyter notebooks, demonstrating methods to detect aspect terms from reviews. Two types of classifiers are explored -- Conditional Random Fields (CRFs) and Random Forest.

Content

1. Data observation.ipynb explores the training dataset in the ./data directory.
2. Prediction.ipynb contains the pipeline of aspect term detector (with CRFs and Random Forest) and theirs precision, recall, and F1.

Getting Started

Installing dependencies

$ export ENV_NAME="_some_name_"
$ conda create --name $ENV_NAME --file requirements.txt
$ source activate $ENV_NAME
$ jupyter notebook
$ python -m spacy download en_core_web_lg