Joint-Embedding-Model

This repository contains code for [ Analysis of Twitter Users’ Lifestyle Choices using Joint Embedding Model ], ICWSM 2021.

Data:

1. Please download the 'data' folder from following link (on request):

[Data]

2. 'data' folder should be kept inside the 'Joint-Embedding-Model/code' folder.

for yoga: data/yoga_user_name_loc_des_mergetweets_yoga_1300_lb.csv file contains 6 columns: name, location, description, text, utype, umotivation

for keto: data/weakly_gt_mergetweets_keto_1300.csv

3. Train and Test data are inside data folder.

data/train.txt

data/test.txt

data/train_keto.txt

data/test_keto.txt

Computing Machine:

Supermicro SuperServer 7046GT-TR

Two Intel Xeon X5690 3.46GHz 6-core processors

48 GB of RAM

Linux operating system

Four NVIDIA GeForce GTX 1060 GPU cards for CUDA/OpenCL programming

Software Packages and libraries:

python 3.6.6

PyTorch 1.1.0

jupiter notebook

pandas

gensim

nltk

nltk.tag

spacy

emoji

sklearn

statsmodels

scipy

matplotlib

numpy

preprocessor

transformers

Run all codes from 'code' directory.

cd code

Construct Yoga User Networks:

1. data/yoga_user_name_loc_des_mergetweets_yoga_1300_lb.csv file) (save in yoga_mergetweets_gt_1300.csv file containing 7 columns: name, location, description, text, utype, umotivation, uid) and @-mention network (save in data/yoga_graph_@mention.txt file)

get_at_mention_network_yoga_1300.ipynb

2. Create user network embeddings using Node2Vec and input graph for this is data/yoga_graph_@mention.txt

Please download Node2Vec from this link:

https://github.com/aditya-grover/node2vec

This will create following embedding:

data/userNetworkEmb.emb

Run Models:

1. Run joint embedding model

BERT_longroberta_des+loc+Net+twt_utype.ipynb

BERT_longroberta_des+loc+Net+twt_umotivation.ipynb

2. Run description only baseline model.

baseline_BERT_description_utype.ipynb

baseline_BERT_description_umotivation.ipynb

3. Run location only baseline model.

baseline_BERT_location_utype.ipynb

baseline_BERT_location_umotivation.ipynb

4. Run tweets only baseline model.

baseline_LongROBERTA_tweet_utype.ipynb

baseline_LongROBERTA_tweet_umotivation.ipynb

5. Run user network only baseline model.

Net_utype.ipynb

Net_umotivation.ipynb

6. Run joint Description and Location (Des + Loc) model.

BERT_des+loc_utype.ipynb

BERT_des+loc_umotivation.ipynb

7. Run joint Description and Network (Des + Net) model.

BERT_des+Net_utype.ipynb

BERT_des+Net_umotivation.ipynb

8. Run joint Description, Location, Tweet (Des + Loc + Twt) model.

BERT_longroberta_des+loc+twt_utype.ipynb

BERT_longroberta_des+loc+twt_umotivation.ipynb

9. Run joint Description, Location, Network (Des + Loc + Net) model.

BERT_des+loc+Net_utype.ipynb

BERT_des+loc+Net_umotivation.ipynb

10. Run fine-tuned BERT model on Description (Description_BERT).

baseline_BERT_finetuned_description_utype_preprocessed.ipynb

baseline_BERT_finetuned_description_umotivation_preprocessed.ipynb

11. Run fine-tuned BERT model on Location (Location_BERT).

baseline_BERT_finetuned_location_utype_preprocessed.ipynb

baseline_BERT_finetuned_location_umotivation_preprocessed.ipynb

12. Run fine-tuned BERT model on Tweets (Tweets_BERT).

baseline_BERT_finetuned_tweet_utype_preprocessed_split.ipynb

baseline_BERT_finetuned_tweet_umotivation_preprocessed_split.ipynb

For Keto diet:

Construct Keto User Networks:

1. Get unique user id of 1300 keto users (input: data/weakly_gt_mergetweets_keto_1300.csv file) (save in data/keto_mergetweets_gt_1300.csv file containing 6 columns: name, location, description, text, utype_gt, uid) and create @-mention network (save in data/keto_graph_@mention.txt file):

get_at_mention_network_keto_1300.ipynb

2. Create keto user network embeddings using Node2Vec and input graph for this is data/keto_graph_@mention.txt

Please download Node2Vec from this link:

https://github.com/aditya-grover/node2vec

This will create following embedding:

data/keto_userNetworkEmb_@mention.emb

Run model:

keto_BERT_longroberta_des+loc+Net+twt_utype_new_usergraph.ipynb

Citation:

If you find the paper useful in your work, please cite:

@inproceedings{islam2021analysis,
  title={Analysis of Twitter Users' Lifestyle Choices using Joint Embedding Model},
  author={Islam, Tunazzina and Goldwasser, Dan},
  booktitle={Proceedings of the International AAAI Conference on Web and Social Media},
  volume={15},
  pages={242--253},
  year={2021}
}