Change Details

We have developed a model to evaluate revisions on Wikidata according to their likelihood to be reverted (T333892) . The model is available for testing on this [[ https://public-paws.wmcloud.org/User:Diego_(WMF)/WikidataRevertRisk/wikidata_ki_example_notebook.ipynb | PAWS notebook ]]. For reporting bugs on the Annotation tool please use this ticket T344016. In this task we are reporting the model evaluation and improvements. To do this we need to: [x] Release an annotation tool for evaluating the model. (Alpha version can be found [[ https://annotool.toolforge.org/projects/6 | here ]]) [x] Launch a labeling campaign [] Expand the model to different edit types (the current just works for claims and description) [] Review results and launch a second campaign [] If needed (based on the evaluation), improve model accuracy

We have developed a model to evaluate revisions on Wikidata according to their likelihood to be reverted (T333892) . The model is available for testing on this [[ https://public-paws.wmcloud.org/User:Diego_(WMF)/WikidataRevertRisk/wikidata_ki_example_notebook.ipynb | PAWS notebook ]]. For reporting bugs on the Annotation tool please use this ticket T344016. In this task we are reporting the model evaluation and improvements. To do this we need to: [x] Release an annotation tool for evaluating the model. (Alpha version can be found [[ https://annotool.toolforge.org/projects/6 | here ]]) [x] Launch a labeling campaign [x] Review results and launch a second campaign [] Expand the model to different edit types (the current just works for claims and description) [] If needed (based on the evaluation), improve model accuracy

We have developed a model to evaluate revisions on Wikidata according to their likelihood to be reverted (T333892) . The model is available for testing on this [[ https://public-paws.wmcloud.org/User:Diego_(WMF)/WikidataRevertRisk/wikidata_ki_example_notebook.ipynb | PAWS notebook ]]. For reporting bugs on the Annotation tool please use this ticket T344016. In this task we are reporting the model evaluation and improvements. To do this we need to: [x] Release an annotation tool for evaluating the model. (Alpha version can be found [[ https://annotool.toolforge.org/projects/6 | here ]]) [x] Launch a labeling campaign [x] Review results and launch a second campaign [] Expand the model to different edit types (the current just works for claims and description) [] Review results and launch a second campaign [] If needed (based on the evaluation), improve model accuracy