Spaces:
Running
Running
Update README.md
Browse files
README.md
CHANGED
@@ -1,14 +1,53 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
---
|
13 |
|
14 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# π Fake News Detector
|
2 |
+
|
3 |
+
**Detects Fake News using an ensemble of 3 Models (Naive Bayes, Logistic Regression, and GloVe-based embeddings)**
|
4 |
+
|
5 |
+
---
|
6 |
+
|
7 |
+
## π¨ Important Disclaimer
|
8 |
+
|
9 |
+
> β οΈ This project is built purely for **educational and experimental purposes** to explore basic Natural Language Processing (NLP) and Machine Learning (ML) techniques.
|
10 |
+
>
|
11 |
+
> β It is **not suitable for real-world fact-checking or decision-making**.
|
12 |
+
>
|
13 |
+
> The models used are simple, non-contextual, and cannot understand language nuances or factual correctness. Misusing this tool for serious analysis may lead to incorrect or harmful conclusions.
|
14 |
+
>
|
15 |
+
> **Please do not trust or rely on the outputs of this demo.** It is meant for **learning only.**
|
16 |
+
|
17 |
---
|
18 |
+
|
19 |
+
## π― Purpose
|
20 |
+
|
21 |
+
This project was created as a part of our research internship as a way to:
|
22 |
+
- Practice building an ensemble model using different NLP approaches
|
23 |
+
- Learn to deploy ML apps with Gradio and Hugging Face Spaces
|
24 |
+
- Experiment with basic text classification on news headlines/articles
|
25 |
+
|
26 |
+
It is **not** a robust or reliable system for determining truth or accuracy in media.
|
27 |
+
|
28 |
---
|
29 |
|
30 |
+
## βοΈ How It Works
|
31 |
+
|
32 |
+
This Fake News Detector uses an ensemble of 3 models:
|
33 |
+
|
34 |
+
1. **Naive Bayes with TF-IDF** β assigns 55% weight
|
35 |
+
2. **Logistic Regression** β assigns 10% weight
|
36 |
+
3. **GloVe Embedding-Based Classifier** β assigns 35% weight
|
37 |
+
|
38 |
+
Each model contributes a score between 0 and 1 indicating the likelihood of the input text being "Real." The final prediction is based on a weighted average.
|
39 |
+
|
40 |
+
---
|
41 |
+
|
42 |
+
## π License & Attribution
|
43 |
+
|
44 |
+
- This project is licensed under the [MIT License](LICENSE)
|
45 |
+
- GloVe embeddings are from [Stanford NLP](https://nlp.stanford.edu/projects/glove/)
|
46 |
+
- This project uses [Gradio](https://gradio.app) for interface deployment
|
47 |
+
|
48 |
+
## π¦ Installation
|
49 |
+
|
50 |
+
```bash
|
51 |
+
pip install -r requirements.txt
|
52 |
+
python app.py
|
53 |
+
|