The ThreadMind interface initiates with an input field where users can submit YouTube or Reddit links for analysis. Alternatively, users can commence analysis by selecting one of the six predefined cards labeled "Get Started."
- Overview: Provides comprehensive metadata, including video and channel statistics as well as descriptions.
- User Interaction: Users can view a snapshot by clicking on the associated image and have the option to watch the video within the ThreadMind interface via the "Watch Video" button.
- Overview: Furnishes metadata including post and subreddit descriptions and statistics.
- User Interaction: Users can click on an accompanying image to enlarge it and can also view any videos embedded within the Reddit post.
- Overview: A succinct summary of comment content is provided in the form of ten bullet points. Keywords are identified and defined.
- Implementation: Leveraging OpenAI's GPT-3.5TURBO API for high-level natural language processing, which directly streams the processed data for real-time analysis.
ThreadMind also allows users to browse through comments and their corresponding replies, thus offering a more nuanced understanding of public sentiment in addition to the analytical data provided.
- Overview: Users can navigate to the original YouTube video or Reddit post.
- User Interaction: Achieved via the "Go to Video" and "Go to Post" buttons respectively.
- Overview: The distribution of sentiment, emotion, and levels of cyberbullying among the comments are visualized.
- Implementation: Uses HighchartsJS for graphical representations. The underlying data is generated by custom-tailored models that are fine-tuned and hosted on Google Cloud Platform, ensuring both accuracy and scalability.
Individuals interested in contributing to the project are encouraged to connect through LinkedIn. You can also checkout the backend code. Thank you :)
The application is live and can be accessed here.
The project is licensed under the MIT License.