| Glorian_WD updated the task description. (Show Details) |
CHANGES TO TASK DESCRIPTION
For instance, suppose there is an item about some person which only contains 2 statements, that aredeveloping the feature, "instance of: human" and "occupation: actor"we'll use a simple weighted sum based on the probability reported via the wbs_propertypairs table to get signal for completeness. Assuming the change on this ticket has been implemented, when we use the API on this item, the API returns properties "instance of", "sex&gender", "date of birth", "place of birth", and "occupation". In this case, the score would be 0.4. This score comes from 2/5, where 2 is the number of suggested properties that the item already contains, and 5 is the total number of items that are suggested by the APIItems with all high probability statements complete should be more likely to *be* complete than items that lack high probability statements.
...
The information about relevant properties allows me to develop a feature for revscoring (https://www.mediawiki.org/wiki/ORES) which signals item completeness. The feature generates completeness score by dividing number of properties that an item contains with number of relevant properties that this specific item should have based on the wbsgetsuggestions API. For instance, suppose there is an item about some person which only contains 2 statements, that aredeveloping the feature, "instance of: human" and "occupation: actor"we'll use a simple weighted sum based on the probability reported via the wbs_propertypairs table to get signal for completeness. Assuming the change on this ticket has been implemented, when we use the API on this item, the API returns properties "instance of", "sex&gender", "date of birth", "place of birth", and "occupation". In this case, the score would be 0.4. This score comes from 2/5, where 2 is the number of suggested properties that the item already contains, and 5 is the total number of items that are suggested by the APIItems with all high probability statements complete should be more likely to *be* complete than items that lack high probability statements.
...
TASK DETAIL
EMAIL PREFERENCES
To: Glorian_WD
Cc: Sjoerddebruin, daniel, aude, WMDE-leszek, hoo, Lydia_Pintscher, Halfak, Glorian_WD, Aklapper, GoranSMilovanovic, Adik2382, Th3d3v1ls, Ramalepe, Liugev6, QZanden, Sethakill, Lewizho99, Maathavan, dg711, Izno, Wikidata-bugs, jayvdb, Anomie, Mbch331, Legoktm
Cc: Sjoerddebruin, daniel, aude, WMDE-leszek, hoo, Lydia_Pintscher, Halfak, Glorian_WD, Aklapper, GoranSMilovanovic, Adik2382, Th3d3v1ls, Ramalepe, Liugev6, QZanden, Sethakill, Lewizho99, Maathavan, dg711, Izno, Wikidata-bugs, jayvdb, Anomie, Mbch331, Legoktm
_______________________________________________ Wikidata-bugs mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/wikidata-bugs
