- By Hanjin Choi, Wesley Bailey, Joo Hwan Yeon, Dong Hwan Kim.
To check our main project details, go take a look at GNN.pdf!
In data preprocessing part_sentiment you might need...
- pip install transformers
In sentiment, I used pretrained transformer model to gain sentiment scores driven from the text reviews. Since I let that args as default, transformers pipeline will take distil BERT_based finetuned model.
Data preprocessing GNN,
I wasn't able to find the data set that includes all the text review, ratings, and categories etc.. So I just build one!
By concatanating with other dataset that includes, categories, ad_methods, time of accesses. Finally we now have 414,805 features of medium sized dataset for our GNN GCMC model.