Upload 20 files

.gitattributes

@@ -0,0 +1,3 @@
+data/raw/fake-and-real-news-2/fake_and_real_news.csv filter=lfs diff=lfs merge=lfs -text
+data/raw/fake-and-real-news/Fake.csv filter=lfs diff=lfs merge=lfs -text
+data/raw/fake-and-real-news/True.csv filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,121 @@
+venv
+models
+
+# Python virtual environments
+venv/
+.venv/
+env/
+.env/
+ENV/
+env.bak/
+venv.bak/
+
+# Python cache and compiled files
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# Jupyter Notebook checkpoints
+.ipynb_checkpoints/
+notebooks/.ipynb_checkpoints/
+*/.ipynb_checkpoints/
+
+# Model files and training outputs
+models/
+*.h5
+*.hdf5
+*.pkl
+*.pickle
+*.joblib
+*.pt
+*.pth
+*.ckpt
+*.safetensors
+wandb/
+tensorboard_logs/
+mlruns/
+
+# Data files (uncomment if you have large datasets)
+# data/
+# *.csv
+# *.json
+# *.parquet
+# *.feather
+
+# Training and evaluation results
+results/
+outputs/
+logs/
+*.log
+
+# Environment variables and configuration
+.env
+.env.local
+.env.*.local
+config.ini
+secrets.json
+
+# IDE and editor files
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+.DS_Store
+Thumbs.db
+.project
+.pydevproject
+
+# Temporary files
+*.tmp
+*.temp
+temp/
+tmp/
+
+# Coverage and testing
+.coverage
+.pytest_cache/
+.tox/
+.nox/
+.coverage.*
+htmlcov/
+.cache
+
+# Documentation builds
+docs/_build/
+.sphinx/
+
+# PyInstaller
+*.manifest
+*.spec
+
+# Unit test / coverage reports
+.coverage
+.pytest_cache/
+cover/
+
+# Rope project settings
+.ropeproject
+
+# Spyder project settings
+.spyderproject
+.spyproject

.streamlit/config.toml

@@ -0,0 +1,23 @@
+[server]
+# Disable file watcher to prevent PyTorch compatibility issues
+fileWatcherType = "none"
+
+# Disable usage stats collection
+gatherUsageStats = false
+
+# Set headless mode for better performance
+headless = true
+
+# Disable CORS protection for local development
+enableCORS = false
+
+[browser]
+# Disable automatic browser opening
+gatherUsageStats = false
+
+[theme]
+# Optional: set a nice theme
+primaryColor = "#1f77b4"
+backgroundColor = "#ffffff"
+secondaryBackgroundColor = "#f0f2f6"
+textColor = "#262730" 

README.md

@@ -0,0 +1,165 @@
+# Fake News Detection Project
+
+A machine learning project that classifies news articles as real or fake using both traditional NLP techniques and advanced transformer models.
+
+## 🎯 Project Overview
+
+This project implements multiple approaches to detect fake news:
+- **Traditional ML**: TF-IDF vectorization with Logistic Regression
+- **Deep Learning**: Fine-tuned BERT model for sequence classification
+
+## 📊 Performance Results
+
+### TF-IDF + Logistic Regression Model
+- **Accuracy**: 98.62%
+- **F1 Score**: 98.67%
+
+#### Detailed Classification Report:
+```
+               precision    recall  f1-score   support
+
+           0       0.98      0.99      0.99      4284  (Real News)
+           1       0.99      0.98      0.99      4696  (Fake News)
+
+    accuracy                           0.99      8980
+   macro avg       0.99      0.99      0.99      8980
+weighted avg       0.99      0.99      0.99      8980
+```
+
+## 📁 Project Structure
+
+```
+FakeNewsDetector/
+├── README.md
+├── requirements.txt
+├── notebooks/
+│   └── FakeNewsClassifier_HuggingFace.ipynb
+├── scripts/
+│   └── train.py
+├── models/
+│   └── bert-fake-news/  (generated after training)
+├── data/
+├── app/
+└── venv/
+```
+
+## 🚀 Quick Start
+
+### 1. Clone and Setup
+```bash
+git clone <repository-url>
+cd FakeNewsDetector
+```
+
+### 2. Create Virtual Environment
+```bash
+python -m venv venv
+
+# Windows PowerShell
+.\venv\Scripts\Activate.ps1
+
+# Windows CMD
+.\venv\Scripts\activate.bat
+
+# Git Bash
+source venv/Scripts/activate
+```
+
+### 3. Install Dependencies
+```bash
+pip install -r requirements.txt
+```
+
+### 4. Launch Jupyter Notebook
+```bash
+jupyter notebook
+```
+
+## 📚 Dataset
+
+The project uses the `mrm8488/fake-news` dataset from Hugging Face, which contains:
+- **Total articles**: ~45,000
+- **Training split**: 80% (~36,000 articles)
+- **Test split**: 20% (~9,000 articles)
+- **Classes**: 
+  - 0: Real News
+  - 1: Fake News
+
+## 🔧 Models Implemented
+
+### 1. TF-IDF + Logistic Regression
+- **Vectorizer**: TF-IDF with 5,000 max features, n-grams (1,2)
+- **Classifier**: Logistic Regression with balanced class weights
+- **Performance**: 98.62% accuracy
+
+### 2. BERT Fine-tuning
+- **Base Model**: `bert-base-uncased`
+- **Training**: 3 epochs with evaluation per epoch
+- **Optimizer**: AdamW with learning rate 2e-5
+- **Batch Size**: 8 per device
+
+## 🛠️ Usage
+
+### Running the Notebook
+1. Ensure your virtual environment is activated
+2. Start Jupyter: `jupyter notebook`
+3. Open `notebooks/FakeNewsClassifier_HuggingFace.ipynb`
+4. Make sure the kernel is set to "venv" or "FakeNewsDetector (venv)"
+5. Run all cells
+
+### Training BERT Model
+```bash
+python scripts/train.py
+```
+
+The trained model will be saved to `models/bert-fake-news/`
+
+## 📋 Requirements
+
+- Python 3.8+
+- pandas
+- scikit-learn
+- datasets (Hugging Face)
+- transformers
+- torch
+- matplotlib
+- seaborn
+- jupyter
+- ipywidgets
+
+## 🎯 Key Features
+
+- **High Accuracy**: Achieves 98.6% accuracy on test set
+- **Multiple Approaches**: Compares traditional ML vs. transformer models
+- **Easy Setup**: Simple virtual environment setup
+- **Comprehensive Analysis**: Includes confusion matrix and detailed metrics
+- **Production Ready**: Trained models can be saved and deployed
+
+## 🔍 Model Analysis
+
+The TF-IDF + Logistic Regression model shows excellent performance:
+- **Balanced Performance**: High precision and recall for both classes
+- **Low False Positives**: 98% precision for fake news detection
+- **Low False Negatives**: 99% recall for real news detection
+- **Robust**: Handles class imbalance well with balanced weights
+
+## 🚀 Future Improvements
+
+- [ ] Implement ensemble methods combining multiple models
+- [ ] Add cross-validation for more robust evaluation
+- [ ] Experiment with other transformer models (RoBERTa, DistilBERT)
+- [ ] Deploy model as a web API
+- [ ] Add real-time news article classification
+- [ ] Implement explainability features (LIME, SHAP)
+
+## 🤝 Contributing
+
+Contributions, issues, and feature requests are welcome! Feel free to check the [issues page](../../issues).
+
+## 📧 Contact
+
+For questions or suggestions, please open an issue or contact the project maintainer.
+
+---
+
+**Note**: This project is for educational and research purposes. Always verify news from multiple reliable sources.

app/app.py

@@ -0,0 +1,36 @@
+import streamlit as st
+from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
+
+# Set page config
+st.set_page_config(page_title="Fake News Detector", page_icon="📰")
+
+# Hugging Face model path (change this to your actual repo ID)
+MODEL_DIR = "ragkasi/bert-fake-news"
+
+@st.cache_resource
+def load_pipeline():
+    tokenizer = AutoTokenizer.from_pretrained(MODEL_DIR)
+    model = AutoModelForSequenceClassification.from_pretrained(MODEL_DIR)
+    return pipeline("text-classification", model=model, tokenizer=tokenizer)
+
+classifier = load_pipeline()
+
+# UI
+st.title("📰 Fake News Detector")
+st.markdown("Enter a news **headline** or **statement**, and this app will predict if it's **real** or **fake**.")
+
+news_input = st.text_area("✏️ News Text", height=150)
+
+if st.button("🔍 Check News"):
+    if news_input.strip():
+        result = classifier(news_input)[0]
+        label = result["label"]
+        score = result["score"]
+
+        # Adjust label display
+        if label == "LABEL_1":
+            st.error(f"🚨 Likely **Fake News** (Confidence: `{score:.2f}`)")
+        else:
+            st.success(f"✅ Likely **Real News** (Confidence: `{score:.2f}`)")
+    else:
+        st.warning("⚠️ Please enter a news statement.")

data/raw/fake-and-real-news-2/fake_and_real_news.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c27a9eba250b78e392310f35876fe47623768d447715d8784850f91936539be
+size 25876225

data/raw/fake-and-real-news/Fake.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bebf8bcfe95678bf2c732bf413a2ce5f621af0102c82bf08083b2e5d3c693d0c
+size 62789876

data/raw/fake-and-real-news/True.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba0844414a65dc6ae7402b8eee5306da24b6b56488d6767135af466c7dcb2775
+size 53582940

data/raw/liar/README

@@ -0,0 +1,41 @@
+LIAR: A BENCHMARK DATASET FOR FAKE NEWS DETECTION
+
+William Yang Wang, "Liar, Liar Pants on Fire": A New Benchmark Dataset for Fake News Detection, to appear in Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (ACL 2017), short paper, Vancouver, BC, Canada, July 30-August 4, ACL.
+=====================================================================
+Description of the TSV format:
+
+Column 1: the ID of the statement ([ID].json).
+Column 2: the label.
+Column 3: the statement.
+Column 4: the subject(s).
+Column 5: the speaker.
+Column 6: the speaker's job title.
+Column 7: the state info.
+Column 8: the party affiliation.
+Column 9-13: the total credit history count, including the current statement.
+9: barely true counts.
+10: false counts.
+11: half true counts.
+12: mostly true counts.
+13: pants on fire counts.
+Column 14: the context (venue / location of the speech or statement).
+
+Note that we do not provide the full-text verdict report in this current version of the dataset,
+but you can use the following command to access the full verdict report and links to the source documents:
+wget http://www.politifact.com//api/v/2/statement/[ID]/?format=json
+
+======================================================================
+The original sources retain the copyright of the data.
+
+Note that there are absolutely no guarantees with this data,
+and we provide this dataset "as is",
+but you are welcome to report the issues of the preliminary version
+of this data.
+
+You are allowed to use this dataset for research purposes only.
+
+For more question about the dataset, please contact:
+William Wang, william@cs.ucsb.edu
+
+v1.0 04/23/2017
+

notebooks/FakeNews_EDA.ipynb

@@ -0,0 +1,177 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "c:\\Users\\super\\Documents\\CSE Projects\\FakeNewsDetector\\venv\\Lib\\site-packages\\tqdm\\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
+      "  from .autonotebook import tqdm as notebook_tqdm\n",
+      "Generating train split: 100%|██████████| 44898/44898 [00:01<00:00, 25984.86 examples/s]\n"
+     ]
+    }
+   ],
+   "source": [
+    "from datasets import load_dataset\n",
+    "\n",
+    "dataset = load_dataset(\"mrm8488/fake-news\")\n",
+    "dataset = dataset['train'].train_test_split(test_size=0.2)\n",
+    "\n",
+    "# Split out sets\n",
+    "train_ds = dataset['train']\n",
+    "test_ds = dataset['test']\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd\n",
+    "\n",
+    "train_df = train_ds.to_pandas()\n",
+    "test_df = test_ds.to_pandas()\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "label\n",
+      "1    18743\n",
+      "0    17175\n",
+      "Name: count, dtype: int64\n"
+     ]
+    }
+   ],
+   "source": [
+    "import matplotlib.pyplot as plt\n",
+    "import seaborn as sns\n",
+    "\n",
+    "sns.countplot(data=train_df, x=\"label\")\n",
+    "plt.title(\"Class Distribution (0 = Fake, 1 = Real)\")\n",
+    "plt.show()\n",
+    "\n",
+    "# Optional: exact counts\n",
+    "print(train_df['label'].value_counts())\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "text     0\n",
+      "label    0\n",
+      "dtype: int64\n",
+      "                                                    text  label\n",
+      "29516  A federal aid package was all set to pass the ...      1\n",
+      "26424  MOSCOW (Reuters) - Russia s lower house of par...      0\n",
+      "12155  The information below is disturbing and should...      1\n",
+      "14098   There are people out there who are giving the...      1\n",
+      "21622  CARACAS (Reuters) - Cuba’s main regional ally,...      0\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(train_df.isnull().sum())\n",
+    "\n",
+    "# You can also visually inspect\n",
+    "print(train_df.sample(5))\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "# Add new column for text length\n",
+    "train_df['text_length'] = train_df['text'].apply(lambda x: len(x.split()))\n",
+    "\n",
+    "# Plot distribution\n",
+    "sns.histplot(train_df['text_length'], bins=50, kde=True)\n",
+    "plt.title(\"Article Length Distribution (in words)\")\n",
+    "plt.xlabel(\"Word Count\")\n",
+    "plt.show()\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Min: 0\n",
+      "Max: 8135\n",
+      "Mean: 404.48017706999275\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(\"Min:\", train_df['text_length'].min())\n",
+    "print(\"Max:\", train_df['text_length'].max())\n",
+    "print(\"Mean:\", train_df['text_length'].mean())\n"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "venv",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.12.2"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

requirements.txt

@@ -0,0 +1,73 @@
+aiohappyeyeballs==2.6.1
+aiohttp==3.12.6
+aiosignal==1.3.2
+altair==5.5.0
+attrs==25.3.0
+blinker==1.9.0
+cachetools==5.5.2
+certifi==2025.4.26
+charset-normalizer==3.4.2
+click==8.2.1
+colorama==0.4.6
+contourpy==1.3.2
+cycler==0.12.1
+datasets==3.6.0
+dill==0.3.8
+filelock==3.18.0
+fonttools==4.58.1
+frozenlist==1.6.0
+fsspec==2025.3.0
+gitdb==4.0.12
+GitPython==3.1.44
+huggingface-hub==0.32.3
+idna==3.10
+Jinja2==3.1.6
+joblib==1.5.1
+jsonschema==4.24.0
+jsonschema-specifications==2025.4.1
+kiwisolver==1.4.8
+MarkupSafe==3.0.2
+matplotlib==3.10.3
+mpmath==1.3.0
+multidict==6.4.4
+multiprocess==0.70.16
+narwhals==1.41.0
+networkx==3.5
+numpy==2.2.6
+packaging==24.2
+pandas==2.2.3
+pillow==11.2.1
+propcache==0.3.1
+protobuf==6.31.1
+pyarrow==20.0.0
+pydeck==0.9.1
+pyparsing==3.2.3
+python-dateutil==2.9.0.post0
+pytz==2025.2
+PyYAML==6.0.2
+referencing==0.36.2
+regex==2024.11.6
+requests==2.32.3
+rpds-py==0.25.1
+safetensors==0.5.3
+scikit-learn==1.6.1
+scipy==1.15.3
+setuptools==80.9.0
+six==1.17.0
+smmap==5.0.2
+streamlit==1.45.1
+sympy==1.14.0
+tenacity==9.1.2
+threadpoolctl==3.6.0
+tokenizers==0.21.1
+toml==0.10.2
+torch==2.7.0
+tornado==6.5.1
+tqdm==4.67.1
+transformers==4.52.4
+typing_extensions==4.13.2
+tzdata==2025.2
+urllib3==2.4.0
+watchdog==6.0.0
+xxhash==3.5.0
+yarl==1.20.0

scripts/evaluate.py

@@ -0,0 +1,29 @@
+from datasets import load_dataset
+from transformers import (AutoTokenizer, AutoModelForSequenceClassification,
+                          TrainingArguments, Trainer)
+# Reload dataset and tokenizer
+dataset = load_dataset("liar")
+dataset = dataset["train"].train_test_split(test_size=0.2, seed=42)
+
+def simplify_label(example):
+    name = dataset["train"].features["label"].names[ example["label"] ]
+    example["label"] = int(name in ["pants‑fire","false","barely‑true"])
+    return example
+
+dataset = dataset.map(simplify_label)
+
+tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
+# Tokenize the text field (can try combining title + text later for improved performance):
+def tokenize(example):
+    return tokenizer(example["statement"], truncation=True, padding="max_length", max_length=128)
+# Tokenize the dataset
+tokenized_dataset = dataset.map(tokenize, batched=True)
+tokenized_dataset.set_format("torch", columns=["input_ids", "attention_mask", "label"])
+# Load model
+model = AutoModelForSequenceClassification.from_pretrained("models/bert-liar-fake-news")
+# Set up Trainer for evaluation
+training_args = TrainingArguments(output_dir="./results", per_device_eval_batch_size=8)
+trainer = Trainer(model=model, args=training_args)
+# Evaluate
+metrics = trainer.evaluate(eval_dataset=tokenized_dataset["test"])
+print(metrics)

scripts/predict.py

@@ -0,0 +1,14 @@
+from transformers import pipeline
+# Load the classifier
+classifier = pipeline("text-classification", model="models/bert-liar-fake-news", tokenizer="bert-base-uncased")
+# Define the predict function
+def predict(text):
+    result = classifier(text)
+    label = result[0]['label']
+    score = result[0]['score']
+    return label, score
+# Example usage
+if __name__ == "__main__":
+    text = input("Enter a statement to evaluate:\n")
+    label, score = predict(text)
+    print(f"Prediction: {label} (Confidence: {score:.2f})")

scripts/train.py

@@ -0,0 +1,54 @@
+from datasets import load_dataset
+from transformers import (AutoTokenizer, AutoModelForSequenceClassification,
+                          TrainingArguments, Trainer)
+# Get
+# dataset['train'] — 80%
+# dataset['test'] — 20%
+dataset = load_dataset("liar")
+dataset = dataset["train"].train_test_split(test_size=0.2, seed=42)
+
+def simplify_label(example):
+    name = dataset["train"].features["label"].names[ example["label"] ]
+    example["label"] = int(name in ["pants‑fire","false","barely‑true"])
+    return example
+
+dataset = dataset.map(simplify_label)
+
+# Load the tokenizer
+tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
+
+# Tokenize the text field (could also try combining title + text later for improved performance):
+def tokenize(example):
+    return tokenizer(example["statement"], truncation=True, padding="max_length", max_length=128)
+
+tokenized_dataset = dataset.map(tokenize, batched=True)
+# Set the format to torch and specify the columns to include
+tokenized_dataset.set_format("torch", columns=["input_ids", "attention_mask", "label"])
+# Load the model
+model = AutoModelForSequenceClassification.from_pretrained("bert-base-uncased", num_labels=2)
+
+# Define the training arguments
+training_args = TrainingArguments(
+    output_dir="./results",
+    eval_strategy="epoch",
+    save_strategy="epoch",
+    num_train_epochs=3,
+    per_device_train_batch_size=8,
+    per_device_eval_batch_size=8,
+    learning_rate=2e-5,
+    weight_decay=0.01,
+)
+# Initialize the Trainer
+trainer = Trainer(
+    model=model,
+    args=training_args,
+    train_dataset=tokenized_dataset["train"],
+    eval_dataset=tokenized_dataset["test"],
+)
+# Train the model
+trainer.train()
+# Evaluate the model
+trainer.save_model("models/bert-liar-fake-news")
+
+
+