Video statistics database

From Tournesol
Revision as of 17:51, 1 May 2021 by imported>Le science4all (Added percentiles)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

This database is a derived database, which aims to respond efficiently to queries on Tournesol.

A video's statistics is tuple < video_id, reliability_score, reliability_percentile, important_score, important_percentile, engaging_score, engaging_percentile, habit_score, habit_percentile, pedagogical_score, pedagogical_percentile, layman_score, layman_percentile, diversity_score, diversity_percentile, backfire_score, backfire_percentile, entertaining_score, entertaining_percentile, n_contributors, n_ratings, pareto_optimal, public_contributors >.

Currently, the list of public_contributors contains a random set of 20 of the public contributors to the video.

Daily update

Every day, the Tournesol learning algorithm computes the Tournesol scores of all videos, and update the database with the results.

Once all the scores are registered in the database, for each quality criterion, we rank the videos according to the criterion, and we convert the rank of a video into its percentile ranking. This percentile is then saved in the video statistics database.

Hourly update

Every hour, the entries n_contributors, n_ratings, pareto_optimal and public_contributors are updated.

To do so, we go through all the video ratings


Upon new rating.

Upon contributor