How it works:

Data aggregation software is working 24/7 to accumulate data from various sources on the web. The algorithm searches for places and emotions.

Every night (1am UTC) a snapshot of the data is analyzed, and a prediction calculated.

Afterwards visualizations are generated in various designs and formats.

Aggregate

(1) Aggregate

Collect, scrape and search data from various live online sources.

Filter combinations of City-name and Emotion.

The scraper currently recognizes 64 emotions and 3213 cities around the world.

preprocess

(2) Pre-Process

Prepare the raw data of each day for later analysis.

Correlate and normalize values, calculate averages and format data.

analyze

(3) Analysis

  • Statistics
  • Heuristics
  • Extrapolation
  • Prediction

Create dataset for later visualization.

visualize

(4) Visualization

  • Rendering of 2D and 3D base layers (clouds and text)
  • Generation of heatmap overlay for map
  • Composition and arrangement of video layers
  • Add on-screen annotation and scroll-text
  • Encoding and live streaming.
The emotion-forecast is free software. The source is available from the artcollider repository: git://rg42.org/tac_amoeba. If you are interested in developing this project further or want to get access to the database of aggregated data, do not hesitate to contact us.