Talk:Wikidata/Notes/Future

Will Wikidata entries be extractable?[edit]

Here's an example that would seem useful to me: consider the sort of county-by-county unemployment data that is visible in File:Poland_unemployment_BK.PNG and similar maps from all over the world. It would be nice if you:

1. Have a central place where the source data (currently cited in the file as an external link to [1]) is tabulated for every year it can be obtained.

2. Can cite a specific statistic from this data, i.e., the most recent data point for a specific county in Poland e.g. Jelenia Góra and the year it was last updated, and include those seamlessly in the article about that county.

3. Can update a "current" SVG map of unemployment in Poland according to color range grouping of the data. This seems to require some way of encoding in your database an automatic means of starting with a flat data file and teaching it how to take that and convert it all the way to grouped data inserted into a SVG text template. Currently the most advanced method I know of to do this is with a spreadsheet and a plain text editor to work with the file!

4. Can create new files on Commons, one for each past year, so that we're not just reporting on the last year ("recentism"). Ideally can spool these automatically into a current video file of all the years for which data is available.

5. Can use the same data to create graphs en masse, e.g. a two-dimensional graph of the unemployment rate in each and every one of the counties of Poland over 12 years of data, and upload these to Commons, so that they can be accessed via a standard template for inclusion in each county's economy section.

6. Can integrate this data with other data sets from Germany, Ukraine, etc. to create larger regional maps, either using it straight or introducing "fudge factors" if need be to account for some difference in how the statistics are calculated (if such a thing happens and a reliable source for appropriate factors can be found)

7. Can develop "best guess" models to refit data from the 50 pre-1999 political divisions of Poland onto the existing schema as best as possible for the graphs, or automatically generate time lapse maps with the appropriate districting for each year and match them all up when creating the video of unemployment change over time.

8. Can recommend bracketing or color schema to produce the most expressive map possible, given not just the data for one year in Poland, but for many years back comparing all surrounding countries, so that we could have consistent unemployment maps/videos for every country in Europe all with the same clear, simple legends.

Note that the idea would be that every one of the graphs, maps, and videos would get automatically updated after a new year's worth of data was entered into Wikidata!

Are you aspiring to do that? Wnt (talk) 20:49, 17 August 2012 (UTC)[reply]

1: That'd be Wikidata.

2: Yes. Inclusion in the article text is still being figured out though.

3: There will be an API and people will be able to write scripts/bots using this API to fill Wikidata with data. And of course they can enter it by hand if it's not too much. As for creation of the SVG: This might or might not happen in phase 3 of the project. If it doesn't there is however nothing stopping anyone from doing that later.

4: We'll not do that in the initial development but should be possible.

5: Possible but why not avoid the creation of the SVG and uploading it to commons and instead visualize the data directly and include that in the article?

6: I can't comment on that yet. It's part of phase 3 and that hasn't been specked out enough yet. We'll keep use-cases like that in mind.

7: Will likely not be done in the initial development.

8: Should be part of the visualization and can probably be done. I can't say if this will be done as part of the initial development.

So to answer your question: Yes but we're focusing on building the building blocks for what you want to do. It might not happen immediately. --Lydia Pintscher (WMDE) (talk) 10:57, 20 August 2012 (UTC)[reply]

I forgot this but maybe looking at the Semantic Results Formats Extension makes things a bit clearer. Especially look at the linked documentation for a list of supported formats. We'll be looking into reusing the formats (and code) it provides. --Lydia Pintscher (WMDE) (talk) 11:24, 21 August 2012 (UTC)[reply]

I should say I'm very impressed with a much more detailed and vastly more optimistic answer than I'd expected here. The last wikilink, though, didn't get me anywhere - did you mean Semantic_MediaWiki#The_Semantic_MediaWiki_extension which leads to [2], or something else? Also I don't know how you'd "visualize the data directly", but it sounds like an appealing feature! Wnt (talk) 19:56, 26 August 2012 (UTC)[reply]

Urgh, sorry - wrong wiki... Here are the correct ones: this and this --Lydia Pintscher (WMDE) (talk) 07:57, 27 August 2012 (UTC)[reply]

Bibliographies[edit]

The main problem with bibliographies is that one item (a work) can contain different editions:

Example item Q25435:

English edition: Social Outsiders in Nazi Germany
1. Non-fiction book. Ed. by Robert Gellately and Nathan Stoltzfus (2001)
2. ISBN 978-0-691-00748-9
Translation: Nazi Almanyasında Toplumdan Dışlananlar
1. Ankara: Phoenix Yayınevi, 2002
2. ISBN 975-6565-43-8
There might be 2nd and 3rd edition etc.

For a machine-readable bibliography we would need sub-items or there would be no possibility to cite one specific edition.

See also: en:Template:Infobox book and de:Vorlage:BibISBN (8076 titles). --Bookman (talk) 20:45, 16 November 2012 (UTC)[reply]

Variants of personal names[edit]

I would be interested in seeing a Wikidata solution to the following issue: quite large numbers of variants in names of persons. This can be important for medieval figures (see e.g. w:User:Charles Matthews/Scholastics). Librarians have traditionally maintained such lists to identify authors. It would be ugly to insist on adding all such variants to Wikipedia articles explicitly; but a collapsed box filled from Wikidata might be a good solution. One might also give an option to create families of redirects in an automated way. Charles Matthews (talk) 09:27, 9 December 2012 (UTC)[reply]

World History Charts[edit]

Provide a GUI that allows users to view historic data displayed in a graphical way, for example with timelines. Depending on active filters different types of data could be juxtaposed, such as persons (with different domains of work or importance being color-coded), historic events (wars, uprisings, revolutions, etc.), creations, inventions, and discoveries, etc. By "zooming in" users would get an adaptive level of detail, similar to how google maps handles geographic data. Such a "synchronoptic history view" isn't a new idea (it's been used in books as well as online here: http://www.hyperhistory.com/chart/history1.html, but has never been attempted with crowd-sourced content. Such a project probably requires relatively little new content. It's mostly about an additional frontend (presentational layer).

Wikiquote[edit]

I'm very active on NL.Wikiquote. I've worked a lot on standardization, quality control and templating. The plans to use Wikidata for Wikiquote certainly sound interesting. How do I get in on that? Whaledad (talk) 14:59, 1 March 2013 (UTC)[reply]

For the moment you just have to wait, as this are plans for the future. For Wikiquotes I think it might be a relative simple change, so my personal guesses: 1) yes, it will be implemented on Wikiquotes and 2) probably before the end of 2013. HenkvD (talk) 19:18, 16 March 2013 (UTC)[reply]

2015 seems more likely, if ever happening. There isn't any plan and Denny mentioned how the planning alone takes several months, let alone the execution. --Nemo 18:11, 5 April 2013 (UTC)[reply]