SachYaPaksh

Reclaiming the truth, one click at a time.

SachYaPaksh is an AI-powered media analysis tool designed to detect and flag bias in real-time. By combining Natural Language Processing (NLP) and a fine-tuned DistilBERT model, the platform helps users distinguish between objective reporting and biased influence.

---

🛠️ Technical Implementation

• Backend: Developed in Python to handle real-time data ingestion.
• Data Pipeline: Custom logic scrapes live news via APIs and structures raw data into JSON format for seamless model ingestion.
• AI Model: Utilizes a Transformer-based architecture with a fine-tuned DistilBERT model for sequence classification.
• Bias Detection: Calculates a "Bias Confidence Score" based on narrative portrayal and sentiment analysis rather than simple keyword matching.

⚡ Core Modules

• FairFeed: A real-time news scanner that identifies and flags skewed headlines and narrative slants in live articles.
• DocuCheck: An inference engine that acts as a "mirror" for content creators, performing an instant pass on input text to identify harmful generalizations.
• TrueView: A visual analysis tool using OCR to extract text from posters and ads to find hidden representation gaps that the human eye often overlooks.

🔮 Roadmap & Vision

We are currently scaling SachYaPaksh toward real-time video analysis and browser extensions. Our goal is to make unbiased media analysis accessible to everyone, ensuring that the stories we consume are balanced and transparent.

With SachYaPaksh, we are reclaiming the truth, one click at a time.

---

Developed by Harshika Saxena and Yajur Sharma.