Multimedia:Meeting in Paris/Notes/Metrics
See also Multimedia:Metrics.
Major issues
What are the major issues related to this topic?
- We have only very limited metrics
Reasons to have metrics:
- data to design or improve MW
- statistic indicators to measure progress
- communication / outreach / relationships with GLAMs
Currently available statistics
What are the existing tools or workarounds used to circumvent these issues?
- http://stats.wikimedia.org/reportcard/
- http://commons.wikimedia.org/wiki/Commons:MIME_type_statistics
- http://toolserver.org/~bryan/stats/
- http://stats.grok.se
- http://wikistics.falsikon.de/
- http://www.trendingtopics.org/
- http://commons.wikimedia.org/wiki/Commons:Template_i18n/Interface_language_statistics
- http://commons.wikimedia.org/wiki/User:Multichill/Categorization_stats
- http://commons.wikimedia.org/wiki/User:Multichill/Commonscat_stats
Other possible solutions
What other solutions could be used to circumvent these issues? (Be bold and creative)
- Count of good/bad images (?)
- Metadata that needs to be filled out?
- High vs low-quality/resolution?
- Internal & external usage?
- Distribution of media files by size (ranges)
- Source: own work / batch upload from partnership / other
- Distribution of uploads per wiki-age of users (ranges)
- File views across all projects (only description pages are available) (and at what size?)
What needs to be developed
- aggregate thumbnail views in hitcount stats
- official storage of hit count data on WMF servers and toolserver (ideally in MySQL table?)
- geoIP agP-lookup based aggregation of country information
- deploy GlobalCheckUsage - add view to toolserver
- Pretty one-page summary of where images by a user/in a category are used.
- Pingback/callback system (spam? don't worry yet)
- most important thing we need pingback for is InstantCommons subscriptions to notify of file updates/deletions
- Referer/usage aggregator
- Aggregator eats live squid logs, every few minutes saves aggregated data into MW database:
- which images
- what thumb sizes
- referrers [pages on wikis / external sites]
- geolookup of viewer IPs - regional breakdown
- Aggregator eats live squid logs, every few minutes saves aggregated data into MW database:
Notes on hit stats aggregation at http://www.mediawiki.org/wiki/Hit_stats_aggregation
Prioritization
What issues should we focus on?
- Pageview and referer aggregators: easy wins
- Global Check Usage stats: easy win