Sannita

Wikidata discussion moved

@Amgine: Well, in this case we're talking about news related to the en:United Kingdom of Great Britain and Northern Ireland, so the category will be uniquely related to that. As Sven said, though, the real deal is with articles themselves. We need to think of some solutions about it. --Sannita (talk) 23:22, 14 November 2013 (UTC)

Well, not necessarily. Does the category refer to the legal-political unit of the state of UK, or more generally as news event occurring withing the geographic limits? To exemplify a news event in which this may come into play, the EU was petitioned to investigate certain territorial and border issues between Spain and Gibraltar in September and October. Gibraltar is an overseas territory of the United Kingdom, independent except in matters of defense and foreign relations. More remotely, Diego Garcia-related stories are as likely belong to Template:Cl the UK, and (perhaps most relevantly) none belong to a Mauritius or Chagossian category.

But to address your primary concern, news organizations do order their articles and we can use their model at least in part. News articles are identifiable by:

• The temporal time of the news event, and the article publication.
• The section of the news publication where the article is published, which is generally topical. (For example, politics, crime, and culture are the three most popular secondary news categorizations, the primary news categorizations are local, national, and international.)
• Featured story streams - that is, multiple interconnected news articles about a single larger news event. For example, an executive or legislative election, a war, or a natural disaster.

To identify a story for interwikis, then, one could use a Julian Day temporal block (a period of time centered on a day of the year, for example, a one week block consisting of the three days prior to the day, the day, and the three days following the day) + a very general topical category + a story id. The story id begins with 0, and a different story in that [temporal block + category] gets the story id 1, and so on. The more difficult item is to identify the temporal block, which should be the UTC day of the news event which is the primary focus of the news article. As an example, a current lead article on en.WN is about an announcement by the Bangladesh cabinet, made about noon Dhaka time (UTC+6) on 11 November. If it were the first article entered in wikidata under its broadest topical category (Education) it should be 313-Education-0 (remembering that julian days number from noon, not midnight.)

The problem with this system is: it still requires a human to clarify collisions so it will not scale.

Under the current model of story development used across the wikinews project collisions will be rare. But stories which are world-wide and rapidly developing - for example Typhoon Haiyan - will require more human intervention, and likely the use of a custom category term, e.g. 313.Typhoon_Haiyan.0. (The total number of broad categories should be limited to approximately 12, and mapped across the languages of the project.) I believe it is very possible to work on the Wikidata site to manage these collisions, but someone like yourself with more involvement there can probably better assess whether management of collisions could be doable in an ongoing manner.

Just some of my thoughts on the issue. - Amgine | t 15:09, 15 November 2013 (UTC)