Tuesday, November 12, 2024

Fellows of the Royal Zoological Society of NZW and .. #ChatGPT

Wikipedia knew in a text about a fellow of the Royal Zoological Society of New South Wales. Unlike many other awards it does not have its own article, there is no category for these fellows, it has a paragraph in the article about the fellows.

Wikidata did not know the award. 

The list of fellows on the RZS website is formatted in a "last name, first name" format. There are too many fellows so converting it by hand is inconvenient. As so many people are enamoured by ChatGPT, I gave it a spin. ChatGPT does NOT process websites for me. So I copy pasted the list and asked it to change the order of the surname and the first name. 

I asked it who had a Wikipedia article. It could not tell me but it gave me a list of fellows who likely have a Wikipedia article. For many of them I added the award in Wikidata and for some fellows  I added a new Wikidata item. For many of them I linked publications and this results in a nice Scholia for the award

It would be really cool when there is a Wikimedia AI that will answer questions like: "for the people in this list change the order of the name and check if these Australian award winners have a Wikipedia article or a Wikidata item". Maybe start with a tool for editors and then open it up to the general public. 

Given that Wikipedia is multilingual, what would be the effect of the data for the answers being all Wikipedias AND Wikidata.. Given that Wikifunctions is language agnostic, why not have functions that are a front end to such a Wikimedia AI?

Thanks,

       GerardM

Saturday, November 09, 2024

The story of African award winning scientists using Wikifunctions

You can find the winners of the Alan Pifer Research award on the English Wikipedia. One of them, the 2011 recipient is Mr Kelly Chibale. there are several ways to be informed about him. There is Scholia and Reasonator, both derive from Wikidata and then there is the Wikipedia article. All four provide information, one is unstructured and exclusively in English. The good news is that parts of it have a structure making it easy for tools to analyse and convert to data. 

A person can read an article, find and add an award not in Wikidata and choose to add the awardees or use "Awarder" to do it with less effort. It is good when it is done but analytical tools could do a better job. There are many tools that produce information in a nice layout like Listeria.. Problem is that it is not maintained by Wikimedia and it is not necessarily multilingual. 

And then there is Wikifunctions. It is developed and maintained by the Wikimedia Foundation. It could do all the things that Listeria does. Having a function that does only list all the honours and awards for someone like Mr Chibale would be great particularly when there is a function that brings to the light all the award winners for any award. An article about an award can be minimalist, and still include stuff that typically goes into an info box.

With functions available like this, it PAYS to engage in Wikifunctions for the specifics of a language for a function. It is impossible to include all awards in any language but with some imagination, we can expose information once the necessary functions are available.

Thanks,

      GerardM

Tuesday, October 29, 2024

the virality of co-authors in urology

Happy birthday Wikidata and, many happy returns.

When you start enriching the data for a Dutch urologist, an academic who published quite a number of scientific papers, obviously there must be many co-authors. Many of them are yet to be identified, at this moment for Jakko A. Nieuwenhuijzen there are some 339 still to be added.

The main consideration is what has the biggest impact. As a colleague of Mr Nieuwenhuijzen is known at Google scholar, adding papers for him brought new publications to Mr Nieuwenhuijzen and many of his co-authors. Enriching data for these co-authors makes the graph more complex.

At some point more precision in the data for a single author is no longer worth the effort. When you then find an other urologist with many papers not yet attributed and many co-authors where Wikidata does not know the gender yet, focus shifts and many more edits make their way into Wikidata.

Many of these co-authors are of the same institute but people from elsewhere find their place in these graphs as well. Many are Dutch but as urology knows many international collaborations this is reflected in the expanding number of co-authors. 

As a topic is developed in this way, it easily results in thousands of edits. As many subject are  researched in this way, the enriched data is there for the world to use. This data is only of value when there is a public. Sharing in the sum of all knowledge has always been what we stand for. Sharing freely and widely generates us a a both public and a future.

Thanks,

      GerardM

Sunday, October 27, 2024

The fallibility of notability

When Wikidata will be split up in a "science" part and "all the rest", scientists who have a Wikipedia article will need to be part of the "rest" as well. This is necessary as all Wikipedia articles have a link to Wikidata because of the "interwiki" mechanism.

It follows that there will be an over abundance of USA scientists and there will  hardly be any scientists of Africa or South America. 

Some data about scientists is likely to be considered to be part of "all the rest" awards for instance. Are these scientists who received an award to be known in two data sets? Some scientists had a career as an athlete.. an other reason for duplication. It is hard enough to maintain the interwiki links and existing duplication within Wikidata, it will become exponentially more difficult when another data set is added.

When the creation of Wikimedia Commons was considered, similar good reasons led to hesitation and prevented us to bite the bullet for quite some time. Commons started with the creation of a Wiki, a MediaWiki patch that showed a picture in a Wikipedia and it then took a long time for most of the duplicate pictures to be only in Commons. It was not technically perfect but it was done perfect in the wiki way.

I hope that we will bite the bullet this time as well. With a new unrestricted wikibase, the old batch jobs can be dusted off and make good for the years of academic data we missed. I pray that Scholia will become functional soon after. 

I will still be able to do my Wikidata thing.. projects like African politicians, Muslim countries and their rulers (past and present).. Awards that can do with an update obviously including science awards.. I will not be bored but maybe I will be working .. maybe not.

Thanks,

       GerardM

Saturday, October 26, 2024

Old soldiers never die, they march in the remembrance parades

As our movement matures, people who were there at the beginning, age. They get other priorities, they get sick, operated upon and as a consequence have a windfall of time to do more work at Wikidata. 

I did a similar job for a dear fellow Wikimedian.. It is now my turn, my chirurg is in this picture and as I add missing co-authors this picture becomes more complex. It will also become more complex when existing co-authors are enriched with new and linked papers.

With Wikipedia there is the promise that even though the information will evolve, all the work people have put in will be there in future and enable people to read/study the subjects each editor cared for.

The data of Wikidata as it is will be split in parts. For the best of reasons but once its structure is broken, the tools that bring structure to the data will be broken as well. The same tools that enable the enrichment of the data will be broken. Much of my Wikimedia legacy will be lost because there will no longer be a public enabled to learn about scholarly works in a Wiki way.

For a few years now this sword of Damocles has hung over Wikidata. As a consequence the potential of Wikidata is not being realised. The data could be so much richer when automated processes bring free knowledge together. References in Wikipedia  indicating later papers and improve its quality. 

As long as I can I will do my Wikidata thing; hope is eternal.

Thanks,

       GerardM


Thursday, March 07, 2024

A Red&Blue approach to Wikipedia references.

Elisabeth Bik is according to her Wikipedia article a "scientific integrity consultant". Her work is often to the detriment of the reputation of scientists and the work they do. Many of the scientists have a Wikipedia article and retracted publications serve as references in Wikipedia articles.

Many more publications are retracted, most if not all are registered at Retraction Watch. It is reasonable to expect that many publications serving as references in a Wikipedia are retracted. Arguments used to achieve a Neutral Point of View based on a retracted publications, are wrong by definition. 

When all references of a Wikipedia are registered in a Red&Blue Wikibase and, when all books with an ISBN and scientific publications with a DOI are ALSO known at Wikidata, it becomes possible to offer a new service. A service providing information about retractions and citations to the publications used as a reference.

Such a service is to be interactive as well.. Just consider: a Wikipedian wants to check the quality of a Wikipedia article. An update button, first checks for retractions and for all citing publications. It then checks for missing data like citations and authors. At the same time new references are added; they are  all processed in the same way.

In the background, all publications will be checked by a batch functionality for updates at Wikidata. Particularly for new retractions, authors who claim a publication.. In this way the information on any topic will be as good as we can make it.

  • scientific publications are retracted and these retractions impact our NPOV
  • publications may be used as a reference in multiple Wikipedias
  • keeping information on sources up to date protects our NPOV
  • making the latest references available to all our Wikipedians ensures an optimal result
So what is not to like? 
Thanks, 
        GerardM

Wednesday, March 06, 2024

Another Red&Blue application; the epidemiologist who wrote the book on "Smoking Kills"

Professor Richard Doll of Oxford is considered one of the best epidemiologists of the 20th century. There are 20 Wikipedias who consider him notable enough for an article yet Wikidata had until now no scientific paper associated with him. That was easily solved by disambiguating "author strings" for Mr Doll. 

With currently 54 publications to his name, none of his books are included. At the Open Library, Mr Doll is known five times and several books were known by these different Mr Dolls. All books have now been attributed to the Mr Doll with id OL1150080A. This identifier is now linked on Wikidata and reading the available books can be read by an international public.

All publications known at Wikidata for Mr Doll are represented in his Scholia. Given that there is much more to explore, this representation will evolve over time. People may add books or publications and additional co-authors may be disambiguated (currently a potential of 159 authors). 

The English Wikipedia has a Scholia template and it is implemented on the Richard Doll article. Functionality like this makes all the effort worth it bringing information to a next level of exposure. It works both ways. Suppose that all references of all Wikipedia articles in any Wikipedia are to be found in Wikidata. All of these references will be known in the Red&Blue Wikibase. All references with an identifier like a DOI or an ISBN can easily be integrated in Wikidata for re-use in other Wiki projects. 

With some additional work, it is even possible to associate references to individual statements and have them known in Wikidata as well. Again this promotes exposure of all the work we do and it promotes re-use in other Wiki projects.
  • Scholia is/could be available as a template on any and all Wikipedias
  • You can read books when available at OpenLibrary
  • Anyone can contribute to the tapestry of information for any scholar
  • References can easily be added in Red&Blue Wikibase
  • These references can be linked to Wikidata making for one stop shopping for updates
So what is not to like? 
Thanks, 
       GerardM