onsdag 11 april 2018

Wikidata has a design problem

    Quality of Wikidata.......

    In Sweden the National Archive are releasing for free 100 million documents from church books etc... ==> we will have excellent primary sources that are good candidates to confirm facts

  • Example sources from the Swedish National Archive confirming facts in Wikidata

i.e. its primary sources. Excellent to be used to confirm birth dates etc.... 

What I can see in Wikidata is that too many sources confirm facts and the quality of the sources can be questioned..: e.g. Q44519#P569

were we have both the birth book and
the internet Speculative Fiction Database confirming the facts....

Question: Any activities to "rank" sources and easy extract sources you trust?...The intention with a loosely coupled system like the semantic web is that YOU should be able to select what sources you trust and use them.....

As Wikidata is designed today

  • everyone can nearly add whatever they want
    • i.e. Internet Speculative Fiction Database is used to confirm a birth date
  • as a reader 
    • its difficult to understand what sources has quality
    • its difficult to select what sources you trust and wants to use.... 
From the example above you get the following sources to confirm the Birth date....
  1. Original birth record link
    1. Quality: Primary source
  2. Original household record link
    1. Quality: Primary source
  3. data.bnf.fr
    1. Quality: ??
  4. Dictionary of Swedish National Biography link
    1. Quality: Very high quality
      1. has a track record of working in 100 years 
      2. have a quality process
      3. has very high trust from experts in the field
  5. Encyclopædia Britannica Online link
    1. Quailty: I guess good
  6. SNAC link
    1. Quality ????
  7. Find A Grave link
    1. Quality: depends
      1. user created material 
      2. without a well defined quality process
      3. never use sources 
        1. the FAG system is designed to avoid adding sources
  8. Web archive link
    1. Quality ????

I would like to see a system that I as a reader can trust and displays the best possible source for the facts.....

Question: Are we moving in that direction?

Östra Ämterviks kyrkoarkiv, Födelse- och dopböcker, SE/VA/13685/C/7 (1833-1860), bildid: C0039805_00103

Household record for Selma Lagerlöf

The next step

  1. Start design criteria for a trusted source
    1. Has a well defined quality process ==> we can link to it and the reader can understand it
    2. Experts in the fields recognize the as having quality
    3. Gather experience in the community of the quality
    4. ??+
  2. Make it easy for the reader to understand the quality of a source
    1. Maybe filter facts per source have links to read about the source
  3. Add ranks to refs in Wikidata as we have for values..... ==> primar sources cab be shown first etc...
  4. .......
I quess Wikicite is a step in this direction but what to do now 

Inga kommentarer:

Skicka en kommentar