Finding Signal in Events & Public Attention

This project explores how real-world events propagate into public awareness by correlating objective event data with human attention signals. The goal is not prediction, but understanding timing, magnitude, and decay of response.

The Question

When a significant real-world event occurs, how — and how quickly — does collective attention respond? Are there measurable differences between events that spark discussion and those that pass quietly?

Data Sources

• USGS Seismic Events: Time-stamped, geolocated earthquake data providing an objective event stream.
• Wikimedia Pageviews: Hourly public web traffic capturing aggregate attention across languages and regions.

Approach

• Ingest raw event and attention data independently, without semantic assumptions.
• Define time-based windows around each event to observe pre- and post-event behavior.
• Establish a statistical baseline and measure deviation rather than relying on keywords or labels.
• Quantify response characteristics such as spike magnitude, lag to peak, and duration.

Why This Matters

This pattern — separating raw signals, defining neutral observation windows, and measuring behavioral response — generalizes well beyond seismic data. The same architecture applies to operational incidents, product launches, infrastructure outages, or any domain where understanding response dynamics is more valuable than classification.

Status

This is an active exploratory project. The emphasis is on clean data contracts, reproducibility, and cost-aware analytics rather than premature modeling or prediction.

Back to Home