#Wikidata - Ranking an #infantry battalion

Many infantry battalions have their Wikipedia article. Obviously an infantry battalion is a military unit and they are part of an army. There are categories with such battalions so it is relatively easy to add them to Wikidata.

Reasonator shows for infantry battalion, this is in effect a "subclass", really nice what the upper layers of the Wikidata ontology looks like. In effect it has become a tool to "walk the tree".
Thanks,
     GerardM

#Wikidata - how many ontologies do we need?

When you check out items in Wikidata, chances are that they have links to one or more external resources. The picture to the right shows how DBpedia is connected to even more external resources. As you may know, DBpedia and Wikidata also have a very tight connection.

In a way, this can be compared to the many connections Wikidata has to the categories in the many Wikipedias, in Commons. All of them have structure of some kind. Information is inferred and acted upon. The category "United States Army Medal of Honor recipients" refers to people who received the Medal of honor.

When Wikidata is linked to external sources, these links have to be there for a reason. It can be to compare information or it can be to add missing information either way. Applications based on the ontology structure of the external sources can apply their logic on Wikidata.

Yes, there may be a purpose for yet another ontology structure in Wikidata. It may allow for implied information in tools native to Wikidata. But there is no point to all of them when we do not make use of them.
Thanks,
     GerardM

#Wikidata handwaving II

Being an academic can be a blessing and a curse. What it implies is learning. This means that an academic has mastered "much of the understanding of certain subjects. One of the biggest challenges of someone who has understanding on an academic level is, being understood by "other" people. To achieve this, a clear understanding combined with simple sentences and words is needed.

This type of language is knows as "Jip en Janneketaal" in the Netherlands. Expressing yourself in this way is as hard as expressing yourself in a foreign language.

The next challenge for an academic as expressed in Wiktionary, is "having no practical importance". The combination of unintelligible language and no practical relevance is an absolute killer in any conversation. This is when helpful people are so important; they point to the W3C when the Wikipedia article about OWL ass-umes too much.

The argument about upper ontologies and classes is for me academic. I understand the notions expressed in the W3C document but they are not implemented as a system in Wikidata. I understand the idea behind the database reports with "constraint violations" but there are no constraints in place and not much is done as a result of the reports.

The "Wikidata generic tree" shows how much effort has gone in building something that resembles an ontology. It has 17,612 entries! There are reports with constraints that people think are important. All this work needs to be championed.

Tools like Reasonator work better when an item is recognised as a human, male or female. There should be tools that help an editor add relevant information. The use of such tools should be obvious, it should be default and most importantly they should be up to date. When they are, everybody will use them. As a result we will get more data and we could have something like an ontology as well.
Thanks,
      GerardM

More of that heady stuff

The blogpost about the "heady stuff of Wikidata" proved interesting. Many people read it and recently there was a lengthy reaction from Mike Bergman. It is quite interesting because he indicates that the work he does may be used for Wikipedia and Wikidata as well.

So have a read of what Mike had to say. <grin> Maybe there will be a follow up </grin> ...

Thanks,

        GerardM

Hi Gerard,

Sorry my comments on this are coming in a bit late; I only recently stumbled upon your thoughtful post. 

For some time, as one of the editors of UMBEL, I have been convinced that much could be done to improve the category structure of Wikipedia. In fact, our most recent formal release of UMBEL has mappings to about half of the content of Wikipedia, most done by hand or using heuristics. Our hope in doing this work is to provide consistent organization, faceting, inference and semantic search over Wikipedia's contents. The extension into Wikidata is only raising the importance of a tractable upper structure. 

Though there are literally hundreds of hours in our effort to date, it is not easily repeatable nor scalable. Further, there are many of the longer-tail concepts in Wikipedia for which heuristics and manual approaches simply are not appropriate. 

Our efforts in this area have led us to look at many alternative mapping approaches, many of which we have tried ourselves and tested internally. 

The best that we encountered that moved in the right direction was Aleksander Pohl's work with OpenCyc and Wikipedia. After about a year of discussions, we have just reached a formal agreement with Aleksander and his team to extend his prior efforts, only focusing on UMBEL as the mapping target this time. (Though the same new approaches may again be extended to OpenCyc.) 

The market will likely tell us if small structures (as you cite), medium structures (such as the 28 K concepts in UMBEL), larger structures (the 300K or so in OpenCyc), or other structures (such as SUMO or whatever), will best fit the bill of better organizing Wikipedia content and making it tractable. We are committed to placing our approach into the public domain soon. 

We should have some results to share on our UMBEL-Wikipedia mappings shortly (weeks, not months). We will do what we can to announce this availability, and we look forward to your and the community's comments and response to this effort. We think we will have a winning approach, but only the market will tell! 

Good luck on your own efforts, and let me know if we can be of any assistance. 

Best, Mike

Even more heady stuff about #Wikidata and ontologies

In an interview you ask questions and the answers are not yours. When you are as lucky as I have been, you get food for thought but not necessarily answers that close an argument. Given that it is my blog and I am commenting on Wikidata anyway, I decided to add some of my own thoughts as well.

Ontologies in their structure are representations of the world. Wikidata is used by over 280 languages and these represent many more cultures. When one upper ontology finds adoption in Wikidata it will not fit well with many of them. A good example is the conversion following the "main type (GND)" "person" controversy; it has been decided that this will become "instance of" "human" in stead. This means that every person will be identified as a specimen of the species Homo sapiens. This may be technically correct but I am sure that many people will feel alienated as a result.

For me it is very important to keep in mind what Wikidata is there for. It is first and foremost a repository of information that is to be used. The considerations of other data repositories are very much secondary to that. One key objective is to use the information as used in info-boxes. The box used for Ronald Reagan serves well to illustrate a few points. It is indicated that he was both a governor of California and a president of the United States. For both these offices, a start and stop date and a predecessor and successor are indicated. In a classic semantic annotation, each of these dates and other office holders would be on their own. As a consequence it would not be possible to tell what date applies to what and who preceded or succeeded to what office. This information can only be given in a relation and qualifiers are the tool that is available.

In all this "rigidity" is important. Ronald was an actor before he was a governor or a president. This rigidity is however dependent on the community of Wikidata; it is and remains a wiki. The logic associated with an actor or a politician should apply both to Arnold and Ronald in equal measure.

When it is indicated that somebody is a "human", it is obvious that he is born, may die and has a sex. These are obvious inferences and they can be suggested as easy and obvious qualifiers and give them a preferred order of presence. As they are easy and obvious, it should be possible to map information consistently to other resources like DBpedia.

Given that explicit links exist between Wikidata and an increasing number of external resources, it seems obvious that people will find a way to compare data. The big advantage for Wikidata is that it has an explicit purpose that will ensure that its data is going to be used. It is going to be used by the biggest possible public because the aim is to share the sum of all knowledge.
Thanks,
       GerardM

More heady stuff about #Wikidata and ontologies

When I asked Emw many questions, I received only three answers. Having an opinion on Wikidata and expressing it is hard. It feels very much like exploring new frontiers. The questions did not go away so I asked them again and, I am mighty pleased that Antoine Isaac was willing to provide me with some answers. Antoine does not do Wikidata, he is a/the scientific coordinator at Europeana.eu. He wrote an email to the Wikidata mailing list that got me interested in asking him my questions. I hope we will collaborate for our GLAM data well with Antoine and Europeana.

I exchanged several emails with Antoine but the answers do stand on their own. I will react in a follow up blog post. I hope you appreciate what Antoine has to say as much as I do.

Thanks,

     GerardM

The Wikipedia article calls upper level ontologies "political" why should Wikidata be interested in any of that?

I wouldn't call them 'political', that's a bit far-stretched. Indeed ULOs embody some abstract considerations on how to represent the world. And as said before, this can backfire as soon as you consider an open web of data, where different representations may well co-exist. Considering a single upper-level ontology as the guiding principle for everything is dangerous. But it is true that they provide valuable bodies of knowledge to re-use, and it may make sense to re-use them for specific domains (e.g. biology, geography) that could be compatible as a whole with the approach of one ULO.

Does it not make sense to group statements together as qualifiers as part of a statement (like it is done for office held [1]) ?

I like qualifiers and dislike them at the same time!On the one end, it's good to have some meta-metadata about the provenance of a statement (who made/endorsed it), or its scope (e.g. the time it applies). In your example, this is the "start date" and "end date". This practices actually fits what is happening in the RDF / Semantic Web area, where a lot of work is being done about Provenance, and many people use quad stores with 'named graphs' instead of just triple stores. 

On the other end, I am anxious about qualifiers being used with other semantics that "here's some info about a statement". In your example, "preceded by" and "succeeded by" are more difficult to interpret in this sense:  (compare with property P580 that mentions explicitly "statement"). I mean, it is possible to interpret your qualifiers as data on statement. But I really feel that people (and you?) will understand it as, say "Te Rata is the predecessor of Korokī Mahuta" and not "The statement 'Te Rata held the office of Maori Monarch' is the predecessor of the statement 'Korokī Mahuta held the office of Maori Monarch'". Which should be the right thing to do (I mean, the one compatible with the "start date" semantics).  

Note that what you told in the other email is the kind of use of qualifiers that would worry me: it is the person who has a birth/death date or a sex, not the statement 'is a person'.Of course one could see the birth and death date to influence of the 'date of validity of the statement "is a person"'. But still we'd be talking about two different things, from a knowledge representation perspective. 

Note also (just for the fun of refering to upper-level ontologies and their dangers) that for some ULOs 'is a person' doesn't have a begin and end date. Being a person is 'rigid' i.e. it must stick to the subject forever. You can't have been a person once and then cease to be a person. Even if you die you're still a person...And don't tell me that rigidity foresees the some ontologies may have Person an an anti-rigid property. This may probably not be the choice made in your favourite ULO. Unless it's one that addresses both reality and beliefs as two possible sides of a same property. But then, good luck re-using it!!!

DBpedia does not have qualifiers, will this impact their ability to use data from Wikidata?

I can't really speak for DBpedia. But I'd say that if qualifiers are used in a way that is both consistent and compatible with the understanding of 'named graphs' in RDF, then they might be interested.

As we map Wikidata items to the content in other repositories, what do we need to compare the data from these repositories

Which repositories are you talking about? Which content? Are 'repositories' knowledge bases, like OCLC or Europeana in the book/artwork domain? Is 'content' 'data'? If yes, then what is required to establish correspondences is hard work looking at the fields and seeing how they correspond. This may be of course alleviated if all (wikidata included) look at what's already happening and try to minimize the risk of coming with data models that are too indiosyncratic. (this is why something like RDF is quite useful!)

When differences in content between repositories are found is there a standard method to harmonise the content

Again assuming a reading of 'repositories' and 'content' as above. I don't think there is a standard method. What you can prey for (and of course as designers of data repositories, we are somehow in the position of making it happen!) is that all repositories keep track of as many unambiguous identifiers they can keep track of (e.g. ISBNs for books) which would help automatic reconciliation. Otherwise make sure the data on the content (where 'content'='the object in the real world') is as complete as possible. For works of art that would mean that comparisons can be made on titles, creators, dates and place of creation, etc.

Some answers about the heady stuff of #Wikidata

I asked Emw questions as a result of his email about the migration away from the "GND main type". I am happy with the answers I received and I hope you will enjoy reading them.
Thanks,
      GerardM

At Wikidata, I contribute to discussions about properties, where I espouse using W3C recommendations and conventions from the wider Semantic Web. I'm also active in discussions about how to model molecular biology data.  Outside of Wikidata, I've been an active contributor to Wikipedia and Commons for several years.

I only had time to answer three of your questions, but I did that much pretty extensively.  The remaining questions are mostly beyond my knowledge and I don't have any well-formed opinion on them.  If you'd like, I can try to answer those questions or others next week.

My answers to your questions:

1) The GND system has been ditched. Can you explain why this is a good thing?

The GND main type property has several major problems. Deprecating that property helps us focus on better solutions for classifying knowledge on Wikidata.

Major issues with P107:

1. With the GND main type property, "person" can mean things well beyond the common understanding of that word.  It can mean things like Coco Chanel -- i.e. 'person' as conventionally understood -- or it can mean a god, literary character, pseudonym, collective pseudonym or spirit.  The standard response to this glaring issue is "'person' is meant to generalize, don't take the term literally". That is not a sufficient solution.  If a classification system for all human knowledge considers Vishnu and Coco Chanel to be both be 'persons', that's a big problem.  Beyond giving users bizarrely unexpected query results, it means properties that should be safe to assume for any given 'person' item simply cannot be.
   
2. Any item that is not a person, place, event, organization or work is classified as a "term", which contains virtually no information.  We need to be able to classify things like gravity, carbon, DNA, cancer, clarinet, Twelver Shia Islam, fashion boot, dog and potato as more than simply "terms".  One sixth of the property is kruft.
   
3. Not even the GND directly uses GND main types.  The GND Ontology has a hierarchical class system and the Deutsche Nationalbibliothek -- which developed it -- uses the lowest-level, most specific GND class available for a subject.  This indicates that the GND senses the GND main types are not appropriate to use as they are with P107.
   
4. The nature of P107 implies that the property is only for the highest level of classification, and that additional properties would be needed for each level in the hierarchy of classification for lower-level types. This would entail lots of unnecessary work to create and update classifications. For example, want to specifically classify Nauru? If property P107 were to persist, then you would need to add something to the effect of "main type: Place" and "subtype: Administrative unit". The problem gets drastically worse for subjects with more levels of classification, like organisms, instruments, molecules, diseases, towns, etc.

The GND system itself -- the GND Ontology -- is not the real problem.  The real problem is that P107 is a "main type" property.  In a project to structure all knowledge -- which Wikidata is -- restricting all items into a small set of types will inevitably lead to many, many classifications that are either A) too broad to be useful or B) simply incorrect.

2.  You sent an email where you asked for attention for what is to be next. Why should there be something next?

Because -- although it is complex -- the world has structure, and classes or types are a useful way to express that structure.  The lopsided debates in the Primary sorting property RFC indicate that so-called "main type" properties (sometimes also called "principal group" or "primary sorting" properties) are a bad idea.  However, that does not mean that the basic notion of grouping things into "types" or "classes" is also a bad idea.

A much better solution for classifying things is to use "type" properties recommended for the Semantic Web by the W3C -- that is, use rdf:type and rdfs:subClassOf. These properties exist in Wikidata as instance of (P31) and subclass of (P279). These properties have been part of W3C recommendations for the Semantic Web for almost a decade. They are fundamental properties used in large controlled vocabularies to structure data into knowledge.  They facilitate classification at an arbitrary granularity.  Together 'instance of' and 'subclass of' can classify all subjects and be used to determine precisely where each subject exists in the hierarchy of knowledge -- or, perhaps -- a collection of hierarchies of knowledge.

Not only do they solve those structural problems of P107 and other "main type" properties, but by being based on W3C recommendations, instance of (P31) and subclass of (P279) also make Wikidata more interoperable with the rest of the Semantic Web.

That said, deciding on properties like P31 and P279 is only the beginning of forming a better way to do classification on Wikidata.  We need a way to map the information in P107 to use P31 and P279.  That's a topic of active discussion on Wikidata.

3.  The GND is a library system then you mention upper ontologies. What is the difference, and how are they practical in the Wikidata context?

The GND (Gemeinsame Normdatei) authority file is used as a library classification system, but it's based on the GND Ontology.  The ontology has a hierarchy of high-level entities and sub-classes.  The P107 property is based on those so called "high-level entities", which were called "main types" in Wikidata as shorthand.  The main GND types are person, place, event, organization, work, term or "undifferentiated person".  These main types are fine as a way to classify items of general interest in a large library, but they're much too small to form a sound basis for a classification system for all human knowledge. 

That's what upper ontologies are for.  An upper ontology is a way to have standard vocabulary about high-level entities in our world.  The idea is to formalize these very general concepts in a way that captures the richness of human language while also being precise enough to be machine-understandable. 

For example, the Suggested Upper Merged Ontology (SUMO) sets a class "entity" as the most general type of thing -- everything is an "entity".  From there, SUMO classifies things in the world as either "physical" or "abstract".  "Physical" things can be "objects" or "processes".  "Abstract" things include so-called "set-classes", "propositions", "quantities" and "attributes".  (More information on SUMO is available in Towards a standard upper ontology.)

There are several other upper ontologies available, like BFO and UMBEL.  I am not an expert in ontologies, and I have not learned enough about each of them to make an informed statement on their advantages and disadvantages.  However, because they seem to offer unifying terminology for different domains of knowledge, upper ontologies strike me as something worth consideration by the Wikidata community.

Dbpedia

I chatted with Duesentrieb the other day. He mentioned dbpedia. I had another look and it is really a great resource. For those that do not know, dbpedia is a community effort to extract structured information from Wikipedia and to make this information available on the Web.

It does a great job and it is different from that other project that deals with structured information, Semantic MediaWiki, in that it does operate by data mining information from Wikipedia while Semantic MediaWiki is an integral part of a MediaWiki project.

The great thing of dbpedia is that it explicitly encourages interlinking. With interlinking data, data that can be found in another resource, becomes available limited by the quality of the interface.

You might ask how OmegaWiki fits into all this. Dbpedia's information is in English while OmegaWiki allows for the representation of information in any language . Both OmegaWiki and dbpedia link to Wikipedia articles and consequently where the two share a link, the information can be mashed together.

In Wikiprotein there is a large amount of medical information available. This information does link to external databases. Given that the medical articles relate to external databases, there is an opportunity to link the data to Wikipedia articles. With Wikipedia articles linked in this way, more specialists will find their way to the Wikipedia medical articles and this in turn will make more enriched information available.

The thing to consider now is how OmegaWiki can benefit from dbpedia.. one of the issues is the difference in license. Then again, dbpedia provides its algorithms and consequently the result is not necessarily the license that dbpedia posts.

Thanks,
GerardM

Upper ontologies

"An upper ontology attempts to create an ontology which describes very general concepts that are the same across all domains". As almost always there is a Wikipedia article about this. Given that the subject is difficult, there is since August 2006 a request to clean up this page .. I have to agree that the subject is interesting and potentially controversial.

For OmegaWiki, an upper ontology is one of those things.. An upper ontology defines the broad strokes, and while it descends downwards more and more aspects may be inherited from higher levels. This means that an upper ontology has practical implications. It also means that we will eventually have in effect an upper ontology by default.

As an upper ontology creates the concepts that are true across domains, Wikis for Professionals will want to define how their ontologies fit into the upper ontology. As they are bound to have overlaps with other domains, there will regularly be found to be in conflicts between domains. As one WfP needs to link into other domains, the question of primacy will raise its ugly head. One WfP was there first, but this other domain is not its competency... A new WfP does have the competency but is disrupts the existing WfP...

An initial selection of an upper ontology will be crucial, its evolution will be exceedingly important when OmegaWiki will is to be bound by the integration into it. Personally I expect that a different model will arise; one where on the one hand great care will be given to the evolution of the upper ontology while on the other hand functionality will be created irrespective of the upper ontology.

Consider, when you know that something is a plant, you can infer all kinds of things about it. It is not really necessary to know how it fits in the greater scheme of things. It would be nice, but it is not required. When a lot of functionality is determined in such a way, there will be the question how this will fit together; it is therefore my prediction that these two forces will eventually find a balance. As OmegaWiki matures, this balance will become increasingly stable.

Thanks,
GerardM