#Wikipedia vs #Wikidata - the George Polk Awards

Some Wikipedians consider Wikidata inferior, so much so that they agitate towards a policy that bans Wikidata in "their" Wikipedia. They are welcome to their opinion.

I do bulk imports from Wikipedia and all the time I suffer the consequences. Some three to four percent of their data is wrong for all kinds of reasons, reasons that are manageable with proper tooling.

The George Polk Award is an award for journalism and it got my attention again because the International Consortium of Investigative Journalists received it for their work on the Panama Papers. I noticed that many people listed who had been awarded the Polk Award did not have articles in Wikipedia, that many of the link in the list of award winners pointed to the wrong person and that many award winners did not even have a "red link".

I am in the process of checking all the links and adding the date for the award. I found many issues among them a civil war general and many others false friends. I am adding items for the people who do not have an English article and, I have to check each of them because several do have articles in other languages. It is a lot of work and it is not as useful as it could be because Wikipedia hates Wikidata and we do not collaborate, we do not work together.

There is a Listeria list of winners and slowly but surely it will contains the information that is similar to the English Wikipedia list article. Similar but not the same;

• the false friends will not be there, 
• there will be no red or black links
• people who won the award twice will be missing

Why do this, why spend so much time on one big list? Well, in this day and age of "fake news" we should celebrate journalism but having all this information in Wikidata allows for all kinds of tools as well. We can check for false friends, we can check if the articles on the award winners include the award but also if there are "winners" who are not known in this list and in the source available for the George Polk winners..

I am not a Wikipedian and truthfully I hate the endless and senseless bickering that is going on. So let me work on the data, make it available to tools. Now you Wikipedians, you may choose not to show Wikidata data in your infoboxes but you will not make your errors go away without collaboration. Yes, you can quote a source but when your data is not in line with what the source states, having a source does not do you good, effectively you provide fake information.

My request to the reasonable people at Wikipedia and Wikidata, let us work together and see how we can improve quality. Lets link wiki links (blue, red and black) to Wikidata and improve the quality of what is on offer first.
Thanks,
       GerardM

#Wikidata Tool - The #Awarder

The Awarder is a tool I use everyday to add people known to have received the award to Wikidata. Its use is straight forward:

• find a list of award winners, a list that includes the person and the year it was conferred
• copy the source text into the awarder
• identify the wiki the data is from
• identify the award by its Wikidata identifier.
• open the results in "quick statements" for processing. 

Easy. When done properly the result is as good as the information from the Wikipedia it came from.

There are a few points. Some lists, like the one on the John Wesley Powell award, have the year on a line and the data is implied for the following text. The results is ten people identified. There are a few red links in there for instance for "George M. Hornberger" and Awarder has identified him so that I can click on a button to find him in Wikidata. As I did not, I added him in Wikidata for later processing. Awarder does not identify organisations as award winners so I had to add the identifier for for instance the "California Department of Transportation". John Galetzka is the award winner for 2016. He is a "black link" so I identified him in the tool with brackets and as a result I could add him as well.

For fifteen award winners it is now known that they won the award. Slowly but surely it adds to the relevance of these people in Wikidata and the missing award winners become easier to identify for the implied notability.

Thanks,

       GerardM

PS thank you Magnus for a great tool

#Wikidata VS #Wikipedia - the issue with input, output

I was told that I should not talk about quality because "on the basis of my work I did not give a good example". Basically I was told to stop what I am doing. As I have written a lot about quality and argued how we can achieve greater quality it is not funny nor is it appreciated but the guy has a point.

With 2,304,191 edits there must be a lot that is wrong in what I have done. No matter how careful I am, the percentage of errors that is to be expected means that with 6% there must be at least some 138,252 errors that I introduced. The problem is that depending on your outlook this is acceptable or it is not. When in stead of me 100 people did the same work, the result would have been the same; together they would have introduced around 138,252 errors as well.

I totally agree that we need to bring our errors down. There are three steps where errors have their origin; input, process and output.

• My input is based on the Wikipedias; their content all have their own issues. They all operate on their own little islands; there is no or little coordinated effort to make the quality of the information we provide a collective ambition.
• My process is based on identifying what I want to work on; typically awards, often the enrichment of data around one person. For tools I mainly use what Magnus provides; they provide superior usability. Reasonator makes Wikidata statements intelligible, it provides superior disambiguation and automated descriptions. Awarder adds both the year and the person who received an award. It allows me to effectively cover a lot of ground. They are the tools I use most, others like PetScan are also invaluable.
• There is too much output I generate and consequently I do not care for individual edits. I justify them all for the process, the routines I follow. I added "Claudia Wills" based on the information in the article of the eponymous award. Like other notable birdwatchers, Mrs Wills does not have her own article and I added her to complement the information on the award.

We share in the sum of knowledge and when the quality of what we provide is to improve, our movement has to become dedicated to the quality of all our information. The typical Wikipedian does mostly care about his or her own project and that is fine; we do not need all of them in an effort to improve our overall quality. The effort I propose can be hidden from view.

A Wikipedia article contains many links; they are blue, red or black. All the blue links are implicitly linked to Wikidata items. Many issues become evident when they can be compared with the links in articles in other Wikipedias or Wikidata. Some Wikis have additional links and they can be mapped to red links and black links. This prevents problems when articles are written with the name suggested in this link.

Once articles on a same subject in many Wikipedias are linked, all kinds of additional functionality become easier; one that is close to my heart is when a new award winner becomes known..
Thanks,
      GerardM

Comparing #Wikipedia using blue, red and black links

There are reasons to compare Wikipedia articles on the same subject in multiple languages. When you just want to read, you may find additional information in another language but as you can imagine, the content should be largely the same. Consequently, the links in an article should go to articles that are about the same topic.

One problem with "blue" links is homonymy. You write a subject in the same but they are not the same; John Doe is one example. Finding these issues, issues that are surprisingly common, can be done by a bot using the Wikidata identifiers for the linked articles.

When there is no article to link to, there is no implicit link to Wikidata. There are two options; we can fake a link by accepting the red or a "black" link as synonymous or we can link a red or a "black" link to Wikidata. The latter is precise and has additional benefits.

When all links are associated with Wikidata items, it is obvious what links in what language are missing or are additional. They are of interest because they may imply potential information to be added to articles or they may point to errors even vandalism. Another benefit is that it helps establish a baseline for a NPOV or neutral point of view without a need to understand the language.
Thanks,
      GerardM

#Wikipedia - Blue, red and black links

Lists in Wikipedia, like this list of award winners of the Tony Kent Strix award on the right exist as blue, red and "black" links. At the moment only an article in English exists about the award and based on past experiences it is likely that other award winners are known in other Wikipedias.

Based on the information in the article, it was easy enough to add the missing information in Wikidata for all the "black links". When you now compare the information in Wikidata with the Wikipedia article, it is feasible to link fixed text to a Wikidata item. This makes it feasible to trigger a warning once a blue link is possible based on new  Wikidata information. In this way a link to Jack Mills is already likely.

When we can compare the information in an article with data in Wikidata, there is an additional way to compare the information and prevent errors and vandalism. Wikidata is after all superior in its use as a tool for disambiguation.
Thanks,
     GerardM