Classification problem of News Report (document) for classes (FAKE, REAL). Try text-related classifiers such as Naive Bayes, MaxEnt, SVM. Use NLTK+SKLearn, NLP Pre-processing, Classifiers and CV-evaluation.
fake_or_real_news_training:
- ID: ID of the tweet
- Title: Title of the news report
- Text: Textual content of the news report
- Label: Target Variable (FAKE, REAL)
- X1, X2 additional fields
fake_or_real_news_test:
- ID, title and text
- Predict Label
Section 1:
- Variable analysis
- Features
- Other insight
- Data Processing
- Drop features
- Label,
- Baseline Models
- Navie Bayes
- MaxEnt (not working -> config issues -> unresolved)
- SVM
Section 2:
- POS Tagging
- POS Model
Section 3:
- Bag of Words Approaches
- Model Optimization
Section 4: Please note this section uses gloVe6b.50.txt file for embedding. Please download: https://nlp.stanford.edu/projects/glove/
- Kears Models
- Seqential NN
- CNN
- LTSM w. Embeddning
Section 5:
- Ensamble
- Final Prediction
- Submission