Data-Driven Trend Intelligence
Scientific Trend Detection
News in Bubbles is a real-time visual trend mapping application. Rather than offering subjective editorial coverage, our technology dynamically analyzes public RSS news feeds and social discussion signals to cluster, measure, and represent the news landscape in a clear, objective bubble format.
E-E-A-T Editorial Commitment: Our platform operates with absolute algorithmic transparency. We link directly to original publishers, bypass clickbait, and offer transparent criteria for score metrics to provide trustworthy news mapping.
1. The Algorithmic Trend Score
Every bubble's visual weight is determined by its Trend Score. This metric reflects a combination of coverage density, conversation interest, and growth rate calculated over the last 24 hours. The calculation operates under three primary vectors:
1. Signal Recency
Recent articles receive a logarithmic weight multiplier. The highest bonus is awarded to stories published within the last 2 hours to capture breaking news first.
2. Source Coverage
We weigh stories based on the number of distinct trusted publishers reporting on them. This reduces single-source noise and biases.
3. Social Pulse
Our algorithm pulls from public discussion signals, checking post engagement and comment velocity to assess organic conversation momentum.
2. Dynamic Story Clustering
Our backend automatically parses thousands of articles every hour. To group related reports into clean, singular bubble topics, we use a Token-based Cosine Similarity algorithm:
- Tokenization: We strip stop-words (e.g., "and", "the", "with") and evaluate meaningful keywords in headlines.
- Similarity Threshold: We apply a strict overlap threshold of 34% similarity. If a new report exceeds this threshold compared to an existing cluster's token pool, it is merged into the cluster.
- AI Synthesis: Once clustered, we call low-latency LLM engines (like Llama 3) to generate a high-readability summary and clean short title without editorial bias.
3. Sourcing Transparency
We restrict our primary index to reputable news domains with established reporting policies, protecting the radar against low-quality aggregators, content farms, and artificial misinformation. Our primary news database connects directly back to the original authors, respecting content rights.
Contact
For research questions, corrections, editorial partnerships, or technical app support, contact us at Aimcol@yahoo.com.