Inspiration

With the overwhelming amount of information available today, it can be challenging to stay informed and distinguish credible news from misinformation. We wanted to create a solution that not only delivers personalized and relevant news content but also ensures the credibility and reliability of the information.

What it does

Khabri is an AI-powered news assistant designed to enhance the news consumption and production experience. It offers personalized news recommendations based on user preferences, and a fake news detection feature to identify misleading information. Additionally, Khabri provides AI-assisted writing tools for journalists, including grammar checks and style suggestions, and can generate concise summaries of lengthy articles, making it easier for users to consume information quickly.

Tech Stack

Frontend: The user interface is built as an Android app using Java/Kotlin. Backend: We used Django to handle the backend, manage the database, and provide RESTful APIs for the app. Machine Learning: We leveraged scikit-learn for developing and deploying machine learning models for news recommendation, fake news detection, and content summarization. Database: PostgreSQL is used for storing user data, news articles, and other relevant information.

Challenges

One of the main challenges was ensuring the accuracy of the personalized news recommendations and fake news detection. Fine-tuning the machine learning models to provide relevant and reliable content required extensive experimentation and testing. Integrating the machine learning models with the Django backend and ensuring seamless communication with the Android app was also a complex task. Additionally, managing and processing real-time data from multiple news sources presented challenges in terms of scalability and performance.

What we learned

Throughout the development of Khabri, we learned a lot about the intricacies of combining machine learning with mobile and web development. We gained valuable insights into data processing, model training, and fine-tuning algorithms for real-world applications.