![]() | A previous campaign was completed in 2015. We're now working on an additional dataset to train ORES how to catch damage and recognize goodfaith in 2016! |
We'll be using WP:Labels to review 6334 randomly sampled edits as "damaging" and/or "good-faith" in order to train classifiers for mw:ORES.
To get started, sign the list below ({{User|You!}}
), then navigate to
Wikipedia:Labels and click the "install the gadget" button. After you've installed the gadget, go back to
Wikipedia:Labels and select the "Edit quality (20k 2016 sample)" campaign. Post on the
talk page if you run into trouble or you want to discuss an edit.
Labeling of 6,334 revisions: 36.1% complete | ||
Computers are very good at crunching numbers however they are terrible at understanding words and the meaning behind them. However, humans are very good at understanding complex things at a glance. Thus hand coding is a common process by which humans add machine readable information ("labels") to observations. We plan to train machine learning algorithms to predict the quality of edits for use in tools like User:ClueBot NG and WP:Huggle.
![]() | A previous campaign was completed in 2015. We're now working on an additional dataset to train ORES how to catch damage and recognize goodfaith in 2016! |
We'll be using WP:Labels to review 6334 randomly sampled edits as "damaging" and/or "good-faith" in order to train classifiers for mw:ORES.
To get started, sign the list below ({{User|You!}}
), then navigate to
Wikipedia:Labels and click the "install the gadget" button. After you've installed the gadget, go back to
Wikipedia:Labels and select the "Edit quality (20k 2016 sample)" campaign. Post on the
talk page if you run into trouble or you want to discuss an edit.
Labeling of 6,334 revisions: 36.1% complete | ||
Computers are very good at crunching numbers however they are terrible at understanding words and the meaning behind them. However, humans are very good at understanding complex things at a glance. Thus hand coding is a common process by which humans add machine readable information ("labels") to observations. We plan to train machine learning algorithms to predict the quality of edits for use in tools like User:ClueBot NG and WP:Huggle.