Implementation of the paper "Visual Sentiment Analysis for Review Images with Item-Oriented and User-Oriented CNN"
Branch: master
Clone or download
Latest commit 5cbec81 Nov 29, 2018
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
.gitignore Updated .gitignore Nov 29, 2018
LICENSE Initial commit Nov 20, 2018
README.md Update README.md Nov 29, 2018
data_generator.py Added code Nov 29, 2018
eval_base.py Added code Nov 29, 2018
eval_factor.py Added code Nov 29, 2018
layers.py Added code Nov 29, 2018
model_base.py Added code Nov 29, 2018
model_factor.py Added code Nov 29, 2018
tensorboard_logging.py Added code Nov 29, 2018
train_base.py Added code Nov 29, 2018
train_factor.py Added code Nov 29, 2018

README.md

VS-CNN

This is the code for the paper:

Visual Sentiment Analysis for Review Images with Item-Oriented and User-Oriented CNN
Quoc-Tuan Truong and Hady W. Lauw
Presented at MM 2017

We provide:

If you find the code and data useful in your research, please cite:

@inproceedings{vs-cnn,
  title={Visual sentiment analysis for review images with item-oriented and user-oriented CNN},
  author={Truong, Quoc-Tuan and Lauw, Hady W},
  booktitle={Proceedings of the ACM on Multimedia Conference},
  year={2017},
}

Requirements

Training

Train the base model:

python train_base.py --dataset [user,business] --num_epochs 20 --batch_size 64 --learning_rate 0.0001 --lambda_reg 0.0005

Train the factor model:

python train_factor.py --dataset [user,business] --num_factors 16 --num_epochs 20 --learning_rate 0.0001 --lambda_reg 0.0005

Evaluation

Evaluate the base model:

python eval_base.py --dataset [user,business] --batch_size 64

Evaluate the factor model:

python eval_factor.py --dataset [user,business] --num_factors 16