Wikidata talk:WikiProject Anatomy/Ontology of Anatomy/draft

Wikidata has lots of places where ontology isn't well thought out. Approaching everything at once seems to be to complex. Anatomy seems to me like a neat subset on which we can focus to have a consistent ontological model. I hereby invite you to join the discussion.

ChristianKl (talk) 14:41, 8 July 2016 (UTC) Iwan.Aucamp (talk) 15:13, 22 March 2020 (UTC) Was a bee (talk) 14:48, 23 September 2017 (UTC) Okkn (talk) 02:20, 25 October 2017 (UTC) JS (talk) Heihaheihaha (talk) 12:31, 8 January 2024 (UTC)

Notified participants of WikiProject Anatomy

WikiProject Ontology has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead. ChristianKl ❪✉❫ 10:11, 20 May 2020 (UTC)

Should we have a "anatomical structure type" item in addition to the "anatomical structure"- item

Currently, anatomical structure (Q4936952) is inconsistently treated. Sometimes we treat treat it as a metaclass and sometimes as a normal class that can be subclassed by other normal classes. Do we want a "anatomical structure type" for the metaclass and a anatomical structure (Q4936952) for the normal class? If so, is "anatomical structure type" the best name or can we find a better name? ChristianKl ❪✉❫ 10:15, 20 May 2020 (UTC)

@ChristianKl: An important question to consider is, when will something be subclass of (P279)anatomical structure and not instance of (P31)anatomical structure type? If something will always be both it may be a reason to only have one. If we keep using it as metaclass (i.e. with hamstring muscles (Q196002)instance of (P31)anatomical structure (Q4936952)) then the name should change to "anatomical structure type". See Talk:Q4936952#How_about_"anatomical_structure_type"_for_english_label @Infovarius:. Iwan.Aucamp (talk) 17:29, 26 May 2020 (UTC)

I think there's value for consistently translating the is_a relationship in Uberon/FMA to mean subclass_of and treat everything that's in Uberon/FMA as being normal classes. ChristianKl ❪✉❫ 19:26, 26 May 2020 (UTC)

Should all anatomical first-order metaclasses by named X type per convention?

Currently, the existing clear second-order class (Q24017414) that we have in cell type (Q189118) has a name that adds -type to the name of it's is metaclass for (P8225). Is this a pattern that we want for all second-order class (Q24017414) in anatomy? ChristianKl ❪✉❫ 10:19, 20 May 2020 (UTC)

•  Strong support I think yes. Iwan.Aucamp (talk) 17:44, 26 May 2020 (UTC)

How should we determine what gets metaclasses?

My initial idea is that all those classes that have properties for this type (P1963) that doesn't applies to their superclasses should have a metaclass. That means that given that we have muscle origin (P3490), muscular system (Q7060553) would get a "muscle type" metaclass while biceps brachii (Q201363) doesn't get a "biceps brachii type" metaclass because there are no properties for this type (P1963) claims that are special to left biceps brachii (Q66526951), right biceps brachii (Q66526950) and human musculus biceps brachii (Q25950351). left biceps brachii (Q66526951), right biceps brachii (Q66526950) and human musculus biceps brachii (Q25950351) can just state that they are instance of (P31) "muscle type". ChristianKl ❪✉❫ 15:18, 22 May 2020 (UTC)

What should we list on the page that describes the ontology for anatomy?

I intuition is that it's helpful to have an amount of metaclasses in amatomy that's small enough that events that add a new metaclass or remove one can be picked up by watching the Ontology of Anatomy. An automated list neatly does the job. I'm less certain whether the list should contain other information as well. ChristianKl ❪✉❫ 13:36, 24 May 2020 (UTC)

Distinction between terms and the entities that are referred by the terms

Plenty of organs and bones are tagged with instance of (P31) medical term (Q52193405) (a lot of the statements are added by So9q. From my view, we should clearly distinguish items that are about terms from items that are about the referents of terms. ChristianKl ❪✉❫ 16:08, 25 May 2020 (UTC)

Hi, could you give an example? I have since left that way of classifying terms. I have no idea how to move forward. I'm focusing on lexemes at the moment and I don't know how to "group" the medical terms in a meaningful way so that we can e.g. make a medical dictionary from lexemes poiting to some collection of Q-items.--So9q (talk) 17:51, 9 December 2020 (UTC)

@So9q: https://w.wiki/ptx contains 288 of those examples.

It might be that fiber (Q66559339) would be better as a lexeme. The tree of "General anatomical term" of Foundational Model of Anatomy is actually about terms and not entities.

In most cases making a dictionary however requires looking at words one by one to do a good job and I don't think that batch jobs that create thousands of lexemes at once without individual human attention are likely good. ChristianKl ❪✉❫ 18:03, 9 December 2020 (UTC)

I agree about that to some degree, batch jobs are useful to prevent manual typing, but manual intervention is needed to some extent it seems in practice. Often I end up improving the Q-items that MachtSinn suggest to link to a lexeme simply because the descriptions are sometimes low quality. The lexeme project is not really about creating a dictionary though if I understood correctly although you can generate a dictionary from the collection of lexemes and the Q-items they refer to.--So9q (talk) 18:48, 9 December 2020 (UTC)

I myself haven't done much with lexemes, so I don't really have strong opinions about how to best create new ones. ChristianKl ❪✉❫ 00:42, 10 December 2020 (UTC)

Should we add maintained by WikiProject (P6104)WikiProject Anatomy (Q8487304) to all ontology items like cell (Q7868)?

It seems to me like having cell (Q7868)maintained by WikiProject (P6104)WikiProject Anatomy (Q8487304) makes it easier to know where to go for help or suggestions, also keps finding items that matter to a project. Iwan.Aucamp (talk) 07:43, 28 May 2020 (UTC)

• I think adding it to all the metaclasses and targets of is metaclass for (P8225) makes sense. That makes it discoverable enough while still keeping the amount of claim to a manageable level. ChristianKl ❪✉❫ 10:24, 29 May 2020 (UTC)

Tobias1984
Doc James
Bluerasberry
Gambo7
Daniel Mietchen
Andrew Su
Andrux
Pavel Dušek
Mvolz
User:Jtuom
Chris Mungall
ChristianKl
Gstupp
Sintakso
علاء
Adert
CFCF
Jtuom
Drchriswilliams
Okkn
CAPTAIN RAJU
LeadSongDog
Ozzie10aaaa
Marsupium
Netha Hussain
Abhijeet Safai
Seppi333
Shani Evenstein
Csisc
Morgankevinj
TiagoLubiana
ZI Jony
Antoine2711
JustScienceJS
Scossin
Josegustavomartins
Zeromonk
The Anome
Kasyap
JMagalhães

Notified participants of WikiProject Medicine do more people have input over the points raised on this page? ChristianKl ❪✉❫ 10:44, 3 June 2020 (UTC)

Review of initial metaclass items

I created a few initial metaclass items according to the above proposed principles. https://www.wikidata.org/wiki/Wikidata:WikiProject_Anatomy/Ontology_of_Anatomy/draft has a list of them. Does anybody have any objections or other feedback? Processwise I think it's good to first specify those metaclass and then in a next step go through the act of adding the new classes to all existing items of concern.

ChristianKl (talk) 14:41, 8 July 2016 (UTC) Iwan.Aucamp (talk) 15:13, 22 March 2020 (UTC) Was a bee (talk) 14:48, 23 September 2017 (UTC) Okkn (talk) 02:20, 25 October 2017 (UTC) JS (talk) Heihaheihaha (talk) 12:31, 8 January 2024 (UTC)

Notified participants of WikiProject Anatomy ChristianKl ❪✉❫ 16:30, 4 December 2020 (UTC)

@ChristianKl: Thanks for the great job. I think it would be better to have metaclasses corresponding to tissue (Q40397) (ex. epithelium (Q41301)), lymph node (Q170758) (ex. axillary lymph node (Q340695)), and brain region (Q1620186) (ex. hippocampus (Q48360)) as well. --Okkn (talk) 13:10, 5 December 2020 (UTC)

I forgot bone (Q265868). By the way, to be honest, I don't see the need for physical anatomical entity type (Q103820425). Any thoughts? --Okkn (talk) 13:23, 5 December 2020 (UTC)

I added items for the classes you suggested and I do expect that we need a bunch of additional metaclasses in the future. ChristianKl ❪✉❫ 14:05, 5 December 2020 (UTC)

The question of whether we need physical anatomical entity type (Q103820425) is one about what criteria we have for what gets it's own metaclass. I went with the principle of grouping that add properties for this type (P1963) that apply to all the items down the tree getting their own metaclass. anatomical entity (Q27043950) seems to be broad enough that it encompases individual members of species like a dog but a dog has no anatomical location. anatomical relation (Q66536035) is also an anatomical entity (Q27043950) without anatomical location. On the other hand there are items like apex of tooth (Q66559458) that subclass physical anatomical entity (Q66394244) but don't subclass anatomical structure (Q4936952). ChristianKl ❪✉❫ 14:12, 5 December 2020 (UTC)

Ok, I understood your explanation well. Although it is not the scope of this WikiProject, I think we have a similar problem with disease (Q12136), biological process (Q2996394), protein (Q8054), gene (Q7187), etc... --Okkn (talk) 14:39, 5 December 2020 (UTC)

Yes, our ontology isn't clear at many places in Wikidata.

At the moment my idea is to spent a bit additional time clarifying the anatomy metaclass items. Then we can get a bit more attention to them. If that goes well we convert the established anatomy items towards the new model. If that goes well we can do the same thing for other parts of Wikidata as well. ChristianKl ❪✉❫ 14:51, 5 December 2020 (UTC)

Is

⟨ anatomical structure (Q4936952)    ⟩ instance of (P31) ⟨ anatomical structure type (Q103813670)    ⟩

or

⟨ cell (Q7868)    ⟩ instance of (P31) ⟨ cell type (Q189118)    ⟩

invalid? Looks good and convenient to me. (I saw your edit: https://www.wikidata.org/w/index.php?title=Q7868&diff=1317392722&oldid=1308334625) --Okkn (talk) 15:31, 5 December 2020 (UTC)

Yes,

⟨ anatomical structure (Q4936952)    ⟩ instance of (P31) ⟨ anatomical structure type (Q103813670)    ⟩

is invalid because

⟨ anatomical structure (Q4936952)    ⟩ subclass of (P279) ⟨ anatomical structure type (Q103813670)    ⟩

isn't true. (see is metaclass for (P8225): all instances of the subject are subclasses of the object)

Besides that definitional problem of is metaclass for (P8225), muscle (Q7365) for example has no muscle insertion (P3491) and muscle origin (P3490) and I consider it desireable to define the metaclass of muscle (Q7365) in a way that all instances have muscle insertion (P3491) and muscle origin (P3490). This allows the instance of (P31) to specify meaning that isn't already specified by subclass of (P279) muscle (Q7365).

⟨ ocular muscle (Q103845612)    ⟩ instance of (P31) ⟨ muscle organ type (Q103813440)    ⟩

also isn't valid while

⟨ orbitalis muscle (Q6592857)    ⟩ instance of (P31) ⟨ muscle organ type (Q103813440)    ⟩

is. This allows a user who wants to make a query for the kind of things that have muscle insertion (P3491) and muscle origin (P3490) to make a query via instance of (P31). {{ChristianKl ❪✉❫ 15:43, 5 December 2020 (UTC)

Then, how should we handle the items of generic or collective terms (such as cell (Q7868), tissue (Q40397), soft tissue (Q1458306), muscle (Q7365), or tongue muscle (Q3502982))? To avoid misunderstandings and misuses, I think we should prepare appropriate anatomical metaclass (Q94945104) for them. --Okkn (talk) 16:34, 5 December 2020 (UTC)

Anything that doesn't fall under a more specific anatomical metaclass (Q94945104) is a anatomical entity type (Q103812671).

The question is whether we can find a anatomical metaclass (Q94945104) that is more explicit about certain properties not being applicable. We likely need https://www.wikidata.org/wiki/Wikidata:Property_proposal/propri%C3%A9t%C3%A9s_%C3%A0_%C3%A9viter for that. ChristianKl ❪✉❫ 17:09, 5 December 2020 (UTC)

@Okkn: I now created anatomical structure class type (Q103914748) (for items like cell (Q7868))) and localized structure class type (Q103915493) (for items like tongue muscle (Q3502982)). Hopefully once we have https://www.wikidata.org/wiki/Wikidata:Property_proposal/propri%C3%A9t%C3%A9s_%C3%A0_%C3%A9viter it will help with defining those entities more in structured data. Till then I use deprecation a bit. ChristianKl ❪✉❫ 15:15, 6 December 2020 (UTC)

@ChristianKl: "** class type" strategy seems good to me. In other examples, how can we model rib (Q196672)/second rib (Q12068630), carpal bone (Q232377)/scaphoid bone (Q13143), lung (Q7886)/right lung (Q5938041) and pulmonary lobe (Q7259536)/right lung lobe (Q28656735)/upper lobe of right lung (Q28656757)? --Okkn (talk) 03:46, 7 December 2020 (UTC)

Particular and solitary

@Okkn: I started by creating chiral anatomical structure type (Q103982366) after looking at the thesaurus and finding the word "particular". I have to think a bit and do research about whether there are better alternatives. A key question is whether rib (Q196672)/second rib (Q12068630) and lung (Q7886)/right lung (Q5938041) are the same or whether the later needs a different type. ChristianKl ❪✉❫ 11:26, 7 December 2020 (UTC)Ontologically, it seems very useful to have a class for entities that exist once per organism, so I created solitary anatomical structure type (Q103982720) which solves the right lung (Q5938041) problem. ChristianKl ❪✉❫ 11:46, 7 December 2020 (UTC)

@ChristianKl: The following statements seem to be correct, if I understand it correctly:

1. ⟨ solitary anatomical structure type (Q103982720)    ⟩ subclass of (P279) ⟨ anatomical structure type (Q103813670)    ⟩
   and
   ⟨ chiral anatomical structure type (Q103982366)    ⟩ subclass of (P279) ⟨ anatomical structure type (Q103813670)    ⟩
2. ⟨ pulmonary lobe (Q7259536)    ⟩ instance of (P31) ⟨ anatomical structure class type (Q103914748)    ⟩
   ,
   ⟨ right lung lobe (Q28656735)    ⟩ instance of (P31) ⟨ anatomical structure class type (Q103914748)    ⟩
   ,
   ⟨ upper lobe of lung (Q28656741)    ⟩ instance of (P31) ⟨ chiral anatomical structure type (Q103982366)    ⟩
   and
   ⟨ upper lobe of right lung (Q28656757)    ⟩ instance of (P31) ⟨ solitary anatomical structure type (Q103982720)    ⟩
3. ⟨ chiral bone type (Q103982390)    ⟩ subclass of (P279) ⟨ bone organ type (Q103843146)    ⟩
   and
   ⟨ solitary bone type ⟩ subclass of (P279) ⟨ bone organ type (Q103843146)    ⟩
4. ⟨ rib (Q196672)    ⟩ instance of (P31) ⟨ bone class type ⟩
   ,
   ⟨ first rib (Q5454239)    ⟩ instance of (P31) ⟨ chiral bone type (Q103982390)    ⟩
   and
   ⟨ left first rib (Q66503478)    ⟩ instance of (P31) ⟨ solitary bone type ⟩
5. ⟨ clavicle (Q16343)    ⟩ instance of (P31) ⟨ chiral bone type (Q103982390)    ⟩
   and
   ⟨ left clavicle (Q66508023)    ⟩ instance of (P31) ⟨ solitary bone type ⟩
6. ⟨ sternum (Q8481)    ⟩ instance of (P31) ⟨ chiral bone type (Q103982390)    ⟩
   and
   ⟨ sternum (Q8481)    ⟩ instance of (P31) ⟨ solitary bone type ⟩
7. ⟨ carpal bone (Q232377)    ⟩ instance of (P31) ⟨ bone class type ⟩
   ,
   ⟨ scaphoid bone (Q13143)    ⟩ instance of (P31) ⟨ chiral bone type (Q103982390)    ⟩
   and
   ⟨ left scaphoid bone ⟩ instance of (P31) ⟨ solitary bone type ⟩
8. ⟨ lung (Q7886)    ⟩ instance of (P31) ⟨ particular organ type ⟩
   and
   ⟨ right lung (Q5938041)    ⟩ instance of (P31) ⟨ solitary organ type ⟩
9. ⟨ heart (Q1072)    ⟩ instance of (P31) ⟨ particular organ type ⟩
   and
   ⟨ heart (Q1072)    ⟩ instance of (P31) ⟨ solitary organ type ⟩

Could you please tell me your thoughts? --Okkn (talk) 13:28, 7 December 2020 (UTC)

@Okkn: I agree with most. At the moment I'm unsure about

⟨ clavicle (Q16343)    ⟩ instance of (P31) ⟨ chiral bone type (Q103982390)    ⟩

,

⟨ sternum (Q8481)    ⟩ instance of (P31) ⟨ chiral bone type (Q103982390)    ⟩

,

⟨ lung (Q7886)    ⟩ instance of (P31) ⟨ particular organ type ⟩

and

⟨ heart (Q1072)    ⟩ instance of (P31) ⟨ particular organ type ⟩

It depends a bit whether you look from the top down or from the bottom up. Is clavicle (Q16343) more the kind of thing that rib (Q196672) or first rib (Q5454239). From the top down we observe that first rib (Q5454239) has a relationship with second rib (Q12068630) that clavicle (Q16343) doesn't have with anything. If we see chiral bone type (Q103982390) as being about entities that have that kind of relationship with other entities then clavicle (Q16343) isn't included.

On the other hand clavicle (Q16343) is similar to first rib (Q5454239) in that both have two subclasses and rib (Q196672) is dissimilar in that it has a lot more subclasses. We have to think about which of those features in more important. If you have arguments for why you think one is better then the other I'm happy to hear them. ChristianKl ❪✉❫ 13:56, 7 December 2020 (UTC)

Obviously clavicle (Q16343) and first rib (Q5454239) have common characteristics: that is, specific shapes and specific relative position (ignoring the left and right). Therefore, there should be a common metaclass for clavicle (Q16343) and first rib (Q5454239), though I'm not sure the definition of chiral bone type (Q103982390). Similarly, sternum (Q8481) has also a specific shape and a specific relative position, and should have the common metaclass. --Okkn (talk) 15:10, 7 December 2020 (UTC)

Is worth noting that lung (Q7886) was conflated with lung (Q103989602) when the above was written. ChristianKl ❪✉❫ 14:38, 7 December 2020 (UTC)

Why do you think lung (Q7886) is not "lung" but "pair of lungs"? Is kidney (Q9377) also not "kidney" but "pair of kidneys"? Of course, I fully agree that we should distinguish the concepts of "lung" and "pair of lungs". In my sense, however, both lung (Q7886) and kidney (Q9377) themselves should be kept as singular concepts. --Okkn (talk) 15:10, 7 December 2020 (UTC)

I went with the general principle of "when in doubt go with that FMA and Uberon say". I don't have strong opinion about which should be which. If you think it should be the other way around feel free to revert and order it the other way around. ChristianKl ❪✉❫ 15:47, 7 December 2020 (UTC)

X class type

@ChristianKl: If

⟨ anatomical entity class type (Q104007623)    ⟩ subclass of (P279) ⟨ anatomical entity type (Q103812671)    ⟩

is valid,

⟨ anatomical structure class type (Q103914748)    ⟩ subclass of (P279) ⟨ anatomical structure type (Q103813670)    ⟩

should also be valid, right? In other words, "X class type", "particular X type" and "solitary X type" are all equivalently subclasses of "X type". That seems not to contradict my above examples. --Okkn (talk) 02:45, 8 December 2020 (UTC)

• @Okkn: No, anatomical structure type (Q103813670) is partly defined as the kind of entities that have a anatomical location (P927). cell (Q7868) doesn't have that and is a anatomical structure class type (Q103914748) but not a anatomical structure type (Q103813670). ChristianKl ❪✉❫ 04:54, 8 December 2020 (UTC)
  • @ChristianKl: Then localized structure class type (Q103915493), which has a anatomical location (P927)?, is a subclass of anatomical structure type (Q103813670)? --Okkn (talk) 08:00, 9 December 2020 (UTC)
    • @Okkn: anatomical structure type (Q103813670) is intended to be more specific then localized structure class type (Q103915493), it adds additional properties. muscle of anterior compartment of thigh (Q66535588) instance of (P31) anatomical structure type (Q103813670) while localized structure class type (Q103915493) instance of (P31) muscle organ type (Q103813440) subclass of (P279) anatomical structure type (Q103813670). ChristianKl ❪✉❫ 11:02, 9 December 2020 (UTC)

Feedback on Flatness paragraph

• @Okkn: I wrote a paragraph to describe the motivation of having many things on the same level of normal classes at https://www.wikidata.org/wiki/Wikidata:WikiProject_Anatomy/Ontology_of_Anatomy/draft#Flatness:_All_enities_that_have_a_UBERON_ID_(P1554)_are_instances_of_a_second-order_class_(Q24017414) . Do you think the paragraph is helpful in the current version? Do you see ways it could be improved? ChristianKl ❪✉❫ 13:00, 10 December 2020 (UTC)
  • I think the explanation is based on concrete examples and is easy to understand. Maybe it's because I'm not good at English, but I don't know the meaning of the word "flatness" in this place... By the way, I fear that the second sentence "The question of whether human brain (Q492038) instance of (P31) brain (Q1073) is a harder one." may easily mislead beginners into thinking that human brain (Q492038)instance of (P31)brain (Q1073) is true. I think it would be better to clearly state that it is considered incorrect. Or something like "However, it would be difficult to point out whether human brain (Q492038) instance of (P31) brain (Q1073) is wrong or not."? --Okkn (talk) 13:39, 10 December 2020 (UTC)
    • I removed the Flatness term and changed that sentence. ChristianKl ❪✉❫ 14:53, 10 December 2020 (UTC)