This is the repository for the master thesis 'Automated Identification of Information Disorder in Social Media from Multimodal Data'. With the help of NetIdee successfully implemented.
This is the second repository of the master thesis. Its main purpose is the model experiment part. The other repositories are:
- This project is licensed under the GNU General Public License version 3 (GPL v3) - see the GPL.txt file for details.
- This document is distributed under CC-BY-Sharelike-3.0 AT
python3 -m venv ./venv
source venv/bin/activate
pip install --upgrade pip
pip3 install jupyter
pip3 install tensorflow-gpu==2.3.0
pip3 install pandas
pip3 install bert-for-tf2
pip3 install scikit-learn
pip3 install telegram_send
# Configure telegram_send for retrieving status information about training according to:
[Documentation about telegram send](https://pypi.org/project/telegram-send/)
.
├── final_models.py
├── gpl.txt
├── models
│ └── best_models
│ ├── single_meta
│ │ └── weights-improvement-100-0.62.hdf5
│ ├── single_text_comments
│ │ └── weights-improvement-03-0.87.hdf5
│ ├── single_text_title
│ │ └── weights-improvement-02-0.88.hdf5
│ └── single_visual
│ └── weights-improvement-02-0.81.hdf5
├── multi_cased_L-12_H-768_A-12
│ ├── bert_config.json
│ ├── bert_model.ckpt.data-00000-of-00001
│ ├── bert_model.ckpt.index
│ ├── bert_model.ckpt.meta
│ └── vocab.txt
├── README.md
├── requirements.txt
├── text_sequence_analysis.ipynb
├── training_comments_bert_preset.ipynb
├── training_comments_image_preset.ipynb
├── training_image_inceptionv3_preset.ipynb
├── training_image_ResNet101V2_preset.ipynb
├── training_image_resnet_50v2_preset.ipynb
├── training_meta_image_preset.ipynb
├── training_meta_preset.ipynb
├── training_text_bert_preset.ipynb
├── training_text_comments_meta_preset.ipynb
├── training_text_comments_visual_meta_preset.ipynb
├── training_text_image_preset.ipynb
├── training_text_title_comments_meta_preset.ipynb
├── training_text_title_comments_preset.ipynb
├── training_text_title_comments_visual_meta_Add_preset.ipynb
├── training_text_title_comments_visual_meta_Maximum_preset.ipynb
├── training_text_title_comments_visual_meta_preset_bakk.ipynb
├── training_text_title_comments_visual_meta_preset.ipynb
├── training_text_title_comments_visual_preset_Add.ipynb
├── training_text_title_comments_visual_preset.ipynb
├── training_text_title_comments_visual_preset_Maximum.ipynb
├── training_text_title_meta_preset.ipynb
├── training_text_title_visual_meta_preset.ipynb
├── utils
│ ├── callbacks
│ │ ├── callbackUtils.py
│ │ ├── MyCallbacks.py
│ │ ├── MyTelegramCallBack.py
│ │ ├── MyTimeHistoryCallback.py
│ ├── datagenUtils
│ │ ├── datagenUtils.py
│ │ ├── DataSeqMetaModel.py
│ │ ├── DataSeqOneModel_Image.py
│ │ ├── DataSeqThreeModels_text_image_meta_old.py
│ │ ├── DateGenThreeModels.py
│ │ ├── dual_modal
│ │ │ ├── DataSeqImageTitle.py
│ │ │ ├── DataSeqMetaVisual.py
│ │ │ ├── DataSeqTitleComments.py
│ │ │ ├── DataSeqTitleMeta.py
│ │ ├── four_modal
│ │ │ ├── DataSeqFourModels.py
│ │ ├── three_modal
│ │ │ ├── DataSeqTitleCommentsMeta.py
│ │ │ ├── DataSeqTitleCommentsVisual.py
│ │ │ ├── DataSequenceImageCommentsMeta.py
│ │ │ ├── DataSequenceImageTextMeta.py
│ │ └── Untitled.ipynb
│ ├── fileDirUtils
│ │ ├── fileDirUtils.py
│ ├── image_models.py
│ ├── models
│ │ ├── modelUtils.py
│ ├── telegramUtils
│ │ └── telegram_bot.py
│ ├── text_processing.py
│ └── textUtils
│ ├── commentsProcessing.py
│ └── textpreprocessing.py
├── venv
Please see Readme Google Bert Models for further information about the BERT Models.
For this repository please download: Download BERT Model