kyawhtetyang/FakeNewsDetector_BiLSTM
BiLSTM-based fake news detection system supporting binary and multi-class classification, with data preprocessing, feature engineering, model training, and Dockerized workflow.
What's novel
BiLSTM-based fake news detection system supporting binary and multi-class classification, with data preprocessing, feature engineering, model training, and Dockerized workflow.
Code Analysis
19 files read · 5 roundsA binary fake-news classifier using a BiLSTM neural network, served via a Flask REST API with a static HTML frontend for text input and prediction display.
Strengths
Clean separation of concerns across preprocessing, feature engineering, model, evaluation, and pipeline orchestration. The inference artifact caching pattern and CORS handling show practical deployment awareness.
Weaknesses
The model architecture is a textbook BiLSTM with no novel contribution; tests are minimal (only 3 test functions, one of which trains a tiny model on 2 samples) and lack edge-case coverage for the API layer. The Evaluator.evaluate method calls plt.show() which will hang in a headless server environment.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
18
Contributors
1
Files
34
Active weeks
3
Repository
Language
Python
Stars
1
Forks
0
License
MIT