We are automating the process of collecting real-time data of desired topics from various sources, analyze the data and generate insightful content as engaging videos and distribute it to the relevant organizations based on what topic they want to track.
2. Data Collection
Sources
Twitter: Tweets related to given keywords/hashtags.
Reddit: Posts and comments from relevant subreddits.
News Articles: Online news websites and RSS feeds containing the keywords.
LinkedIn: LinkedIn feed and relevant topic discussions.
Methodology
Use web scraping and APIs where available (e.g., Twitter API, Reddit API).
Create our own scrapers, when free APIs are not available.
Keyword-based filtering to fetch relevant data.
3. Sentiment Analysis
Goals
Extract key emotions or topics if possible.
Approach
Use pre-trained sentiment analysis models (e.g., Hugging Face transformers, TextBlob, Vader).
Fine-tune or customize models for domain-specific sentiment detection.
Output
Sentiment-labeled dataset for content generation.
Visualizations or summaries of public mood trends.
4. Text Content Generation
Objective
Generate blog-style textual summaries and insights based on scraped data and sentiment analysis.
Method
Input: Aggregated data + sentiment scores.
Use Large Language Models (LLMs) such as GPT (OpenAI, Hugging Face models) to generate:
Summaries
Opinion pieces
Trend analysis
Features
Tone adaptation (e.g., formal, casual, journalistic).
Inclusion of quotes or excerpts from real posts/articles.
SEO-friendly content generation.
5. Video Creation
Purpose
Create engaging video content to deliver the generated text as engaging news.
Workflow
Convert generated text into a video script.
Use text-to-speech (TTS) engines to produce natural voiceovers.
Integrate with Azure AI avatar services (such as Azure Speech Service with Custom Neural Voice and Azure Video Indexer or Azure Bot Framework with avatars) for realistic lip-sync and facial expressions.
Use text to video creation tools for engaging content delivery.
Add background visuals like relevant images, news tickers, or animations.
Output
Short news-style video clips ready for social media sharing.
.png)
