Wikipedia:Wikipedia Signpost/2010-09-13/Sister projects


 * WereSpielChequers is an editor on the English Wikipedia and occasionally elsewhere. He has been actively involved in various Biography related projects this year and collaborated with Bot writer Merlissimo to launch the Death Anomalies project.

Just over a month ago, The Signpost published a story on the Death Anomalies project, which identifies anomalies where different language Wikipedias disagree as to whether an individual is dead or alive. The Project was started in June, with initially just the German and English language Wikipedias extracting reports of anomalies. Since then, the Latin, Swedish, and Slovenian Wikipedias have joined in, and hundreds of errors have been resolved. When The Signpost covered the project, readers pitched in and the number of anomalies on enwiki was slashed from 447 to 190 in just over a week. EN wiki still has more than a 100 anomalies on Database reports/Living people on EN wiki who are dead on other wikis, with new reports coming in daily. However, most of the backlog is down to differences in the way different projects treat missing people who (if alive) would be more than 100 years old, cross-wiki anomalies stemming from unreferenced articles showing a person as dead, and issues that probably require a native foreign-language speaker to resolve.

In July, only two projects were extracting data from the table, though it queried data from around 70. Subsequently these have been joined by the Swedish Wikipedia which rapidly reduced 94 anomalies to 16, and the Latin wikipedia, which has managed to reduce its anomalies to one. Earlier this month the Slovene Wikipedia became the fifth participating project, and went in a week from requesting a report to having cleared their backlog.

Biographies of living people (BLPs) inevitably need to be updated when the subject dies, so all these reports are expected to be ongoing maintenance tasks. Although the bot is processing data from millions of biographies across different Wikipedias, fewer than a thousand anomalies have been identified so far, relying on Interwiki links and categories that identify biographies as dead or living. Some projects are ineligible for the program because they don't organise their articles in such a way; for example, the Portuguese Wikipedia have lists of people who died in particular years (rather than categories).

In the future, the number of languages from which data is extracted and number of languages requesting reports will hopefully increase; we have 66 Wikipedia language versions including French, Spanish, Japanese, Polish and Russian for whom reports could be extracted almost immediately. Merlissimo (whom Jimbo Wales praised as a "rock star" for his work on the project) has a bot that updates the reports daily, and is willing to produce reports for other projects.

User responses
The Swedish Wikipedia is fertile ground for a project of this kind. After some years of rapid growth in the number articles, attention swung to quality and structure in 2008. Biographic articles were exhaustively categorized by gender in the [northern autumn] of 2008, revealing that there are four male biographies for each female one, and by years of birth and death in 2009. This is also when the category for living people and a WikiProject for living people were started. The "death anomalies" report was set up as a subpage to this WikiProject, named "possibly deceased" people.... The Swedish Wikipedia has also benefited from Check Wikipedia, a daily report of wiki-syntax errors, and would welcome similar projects. (LA2)

Although the Latin Wikipedia (la.wikipedia) uses a language with a long history, a large portion of its articles cover modern topics, including (of course) BLPs.... [Of] about 44,000 articles available in the Latin wikipedia today, about 4300 (roughly ten percent) are BLPs. The death anomalies table adds an extra level of reliability to BLPs on the [English, German, Swedish and Latin Wikipedias]. It is great to see more and more tools are available that permit semantic checks and analyses of information ... the future is not just isolated wikitext articles, but a flexible repository of semantic information. The death anomalies table shows a glimpse of what might be possible in the future, when we will have at our disposal not only (wiki)text but also rich, usefully structured information and data. (UV)

The Slovenian Wikipedia has a relatively large proportion of biographies, of which there are more than 8,000 in BLPs (almost 10% of total article count). Many of those articles have been added semi-automatically and we have a small community of active contributors. Consequently, [many articles] aren't regularly maintained, which is why this tool will certainly prove extremely useful for easing the burden of keeping the content up-to-date. This means less work when the focus shifts from adding content to improving the quality one day, and improved reliability of the work until then. (Yerpo)

The German Wikipedia has more than 340,000 articles about people that include machine-readable data usable by external projects. The local report covers all people (not only living people) and is forwarded to 150 WikiProjects filtered by subject area. The script runs on the toolserver and uses the sun grid engine for efficient resource handling. About 1.9 million interwiki relations are checked every day for creating reports on five Wikipedias. (Merl)