Wikidata:Property proposal/Included in curricula

Curriculum topics[edit]

Originally proposed at Wikidata:Property proposal/Generic

Done: curriculum topics (P8263) (Talk and documentation)

Description	topic or subject included in the curriculum
Data type	Item
Domain	item
Allowed values	items
Example 1	Key Stage 3 (Q6398096): educational level in England and Wales equivalent to lower secondary school (year 7 to year 9) → (start date, end date) → carbon cycle (Q167751) ref (page 9)
Example 2	Key Stage 3 (Q6398096): educational level in England and Wales equivalent to lower secondary school (year 7 to year 9) → (start date, end date) → speed (Q3711325) ref (page 10)
Example 3	Key Stage 3 (Q6398096): educational level in England and Wales equivalent to lower secondary school (year 7 to year 9) → (start date, end date) → flower (Q506) ref (page 6)
Source	external reference URL
Planned use	Map several curricula in Wikidata, we would include qualifiers for start date, end date, grade level/keystage level etc.
Expected completeness	Always incomplete, curricula change over time and new curricula are created.

Motivation[edit]

Mapping curricula on Wikidata will allow us to understand which topics included in curricula have Wikipedia articles and which are still missing. It will also allow us to map curricula to educational resources, both in Wikimedia projects and externally e.g popular educational websites like Khan Academy. I've created an RFC here which outlines more widely a process for connecting resources and curricula. John Cummings (talk) 13:18, 10 April 2020 (UTC)[reply]

Discussion[edit]

Support - NavinoEvans (talk) 17:00, 10 April 2020 (UTC)[reply]
Support - Powerek38 (talk) 07:59, 11 April 2020 (UTC)[reply]
Support - Lirazelf Lirazelf (talk) 11:28, 11 April 2020 (UTC)[reply]
Oppose this in favor of the reverse relation: a curriculum linking to the subjects it covers. Mahir256 (talk) 21:16, 11 April 2020 (UTC)[reply]
Comment aren't there topics that are covered every year in almost every curriculum? cell (Q7868) in biology could be that. Just the level of study changes. --- Jura 07:30, 12 April 2020 (UTC)[reply]
Oppose – It isn't relevant for most people who look at a topic which curricula they belong to. It is more relevant for people looking at the curricula, so the reverse property would be better if there isn't already a usable property for that. --Dipsacus fullonum (talk) 08:15, 12 April 2020 (UTC)[reply]
Comment @Mahir256: and @Dipsacus fullonum:, to be clear, you are not oppose to the idea, you would just prefer the reverse property? So the data would be included on the items each the curriculas and the property would be called something like 'subjects in curricula'? Thanks, --John Cummings (talk) 15:15, 13 April 2020 (UTC)[reply]
- @John Cummings: Yes, exactly. I fear that some topics can be filled with hundreds of links to curricula making them hard to overview. But for a reverse property I think it will be an advantage to be able to see all topics at once. --Dipsacus fullonum (talk) 20:27, 13 April 2020 (UTC)[reply]
  - Thanks for the explanation @Dipsacus fullonum:, I have now changed the proposal to the inverse property. --John Cummings (talk) 18:50, 14 April 2020 (UTC)[reply]

*  Comment The two users who opposed this property above wanted it to be changed to the inverse property which has now been done. --John Cummings (talk) 12:11, 14 April 2020 (UTC)[reply]

Support This will be huge for items with sitelinks to Wikiversity and Wikibooks. The reverse property, on the other hand, might get unwieldy with hundreds or thousands of connecting items from each curricula. Ainali (talk) 18:58, 13 April 2020 (UTC)[reply]
- You mean 300*12 statements (assuming national curricula) or 30000*12 statements on cell (Q7868) is better than, how many? --- Jura 19:17, 13 April 2020 (UTC)[reply]
  - @Jura1: can you explain more what you mean, unpack it so its easier to understand, why are there 100 times more statements in one version than the other? Thanks, --John Cummings (talk) 22:02, 13 April 2020 (UTC)[reply]
    - Both depend on the hypothesis given above (I don't think you responded to that yet). --- Jura 22:04, 13 April 2020 (UTC)[reply]
      - Hi @Jura1:, I don't understand what you mean at all, could you give some more context and explain in more detail? If you're referring to the topics being covered at every level, the answer is sort of. Subjects tend o get more granular over time e.g primary school you would study cells as a broad concept, where as in higher levels of education you would study specific aspects of cells, different kinds of cells, different processes etc. My main question is why if this is modelled one way do you get 100 times more statements (I hope I'm understanding your point properly). Thanks, --John Cummings (talk) 10:06, 14 April 2020 (UTC)[reply]
        Not all countries (300) have a national curriculum for every level (*12). It could vary from one school/state to another (*100). If you model it from curriculum to topic, you would have 30000*12 items pointing to cell (Q7868). Same (theoretical) number. --- Jura 10:14, 14 April 2020 (UTC)[reply]
        Ok thanks. --John Cummings (talk) 10:56, 14 April 2020 (UTC)[reply]

*  Comment The users who opposed this property proposal wanted it to be changed to the inverse property which has now been done. --John Cummings (talk) 12:11, 14 April 2020 (UTC)[reply]

Question Looking at the revised samples and proposal, wouldn't National Curriculum for England (Q17016944) end up linking to everything? The actual information will be in complicated to edit multiple? qualifiers per statement. --- Jura 08:35, 15 April 2020 (UTC)[reply]
- @Jura1: can you explain more about what you mean? I don't understand what you mean linking to everything. Thanks, John Cummings (talk) 11:48, 15 April 2020 (UTC)[reply]
  - Can someone try to explain it to John? --- Jura 11:51, 15 April 2020 (UTC)[reply]
    - I don't think anyone else will be able to explain your question. --John Cummings (talk) 12:18, 15 April 2020 (UTC)[reply]
Comment can you try to do a sample item with Property:P369? It might be easier to see if the approach is feasible and, if it works, the statements can be converted to the new property. --- Jura 11:55, 15 April 2020 (UTC)[reply]
- Mapping a whole curriculum will take a lot of time, what is not clear from the examples? --John Cummings (talk) 12:18, 15 April 2020 (UTC)[reply]
  - See below. It's not clear whether the approach is feasible. --- Jura 13:07, 17 April 2020 (UTC)[reply]
@John Cummings, Jura1: I believe Jura's point is that National Curriculum for England (Q17016944) covers all topics and stages, it would probably be better to divide it up into parts (for example the "programmes of study" listed here and then by stage, or vice versa?) with an item for each, and then point those to the individual topic items via this property? Otherwise you'll end up with thousands of statements for this property on the one item (National Curriculum for England (Q17016944) here) and items with thousands of statements become unmanageable within Wikidata. ArthurPSmith (talk) 17:48, 15 April 2020 (UTC)[reply]
- Thanks very much @ArthurPSmith: for the explanation, are there any other similarly complex topics which have already been imported into Wikidata that we could use to plan how to model curricula? One thing that might make it easier for managing curricula items is whilst curricula are complex they tend to change comparatively rarely, at maximum once per year.--John Cummings (talk) 20:36, 15 April 2020 (UTC)[reply]
  - @John Cummings: I suppose comparable things in complexity would be the various databases, which we usually import by linking their ID's using a custom property... so maybe not comparable, though by that analogy if there were such ID's for these curricula, the link ought to be in the original direction rather than the new reversed one! ArthurPSmith (talk) 20:50, 15 April 2020 (UTC)[reply]
    - Thanks very much, I think some curricula might have IDs although I would guess that this would mean needing one additional ID property per curricula rather than one overall. --John Cummings (talk) 21:33, 15 April 2020 (UTC)[reply]
Comment the perspective of the Education Team at the Wikimedia Foundation is included here

Just one more note: It is important to remember that curricula around the world are not standardized. The only standardization for analysis has come from UNESCO (the ICSE). Names of levels of study and fields of study are examples of what has been standardized by ICSE, but these are not included in local curricula documents. So it could be that we also need to request properties for local education level identifiers and the ICSE levels.NSaad (WMF) (talk) 12:00, 17 April 2020 (UTC)[reply]

Comment to provide a better usecase, maybe you could pick 20 videos from Khan Academy and present how they would be mapped to the UK curriculum and Wikipedia articles. --- Jura 12:37, 17 April 2020 (UTC)[reply]
- The property proposed here does not link to educational resources, only to items curricula. On the left hand side of this page are subjects on Khan Academy that would link to Wikidata items (one or more). I'm unsure if we would need a new property to link from Wikidata items to educational resources, that would be a separate conversation and property proposal. --John Cummings (talk) 18:31, 21 April 2020 (UTC)[reply]
  - Above you explain that "It will also allow us to map curricula to educational resources,". Can you illustrate this with 20 videos from Khan Academy. Maybe a short table above would be helpful. "cell" -> "national curriculum" isn't really illustrating anything. --- Jura 19:01, 21 April 2020 (UTC)[reply]

I've done as many as I'm comfortable with being certain are correct


Khan Academy lesson	Wikidata items
Introduction to algebra	algebra (Q3968)
Solving basic equations & inequalities (one variable, linear)	equation (Q11345) and inequality (Q28113351)
Linear equations, functions, & graphs	linear equation (Q484637), function (Q11348) and graph (Q141488)
Sequences
System of equations
Two-variable inequalities
Functions	function (Q11348)
Absolute value equations, functions, & inequalities
Quadratic equations & functions	quadratic equation (Q41299) and function (Q11348)
Polynomial expressions, equations, & functions	polynomial expression (Q44173099), equation (Q11345) and function (Q11348)
Exponential & logarithmic functions
Radical equations & functions
Rational expressions, equations, & functions
Trigonometric functions
Algebraic modelling
Complex numbers
Conic sections	conic section (Q124255)
Series & induction
Vectors
Matrices	matrix (Q44337)

Some things that have come out of doing this mapping:

I don't understand the subject well so was not able to map all the items, this highlights the need for people who are knowledgeable in the subject and have experience in teaching to do some of the matching.
Many lessons are based on the relationship between different topics and how they work together and so link to more than one Wikidata item.
I think some concepts are missing from Wikidata or not quite granular enough, mapping educational resources could highlight gaps in Wikidata topics.

Thanks

--John Cummings (talk) 12:27, 29 April 2020 (UTC)[reply]

Support in the format of matrix (Q44337) "in curriculum of" National Curriculum for England (Q17016944) with qualifiers that describe it -- I don't think it would be feasible for any one curriculum item to have thousands of different topics -- it actually makes it useless. The end goal of this could be to do as John is describing above, allow education databases to map their tagging data to Wikidata concepts, and deliver curriculum-level topics at the right level. Reading through the proposal, I think folks have a lot of different competing ideas about the model for this -- I think we may want to refresh with a better data modelling examples, Sadads (talk) 13:30, 4 May 2020 (UTC)[reply]
Support Sounds useful to me. Richard Nevell (talk) 17:14, 4 May 2020 (UTC)[reply]

Comment I think the table is an interesting start. To build a more complete sample, it needn't be from a single topic. One could pick a series of videos one understands, map them to items (or create new ones). The question is then how to link that appropriately to the relevant parts of a curriculum (a third column in the table above).
Just linking an item for a curriculum that appears to cover most levels of education in England and covers all topics might not help much. Also, references should point to specific pages/sections.
The risk here is that if this proposal isn't granular enough, we end up with items that are too general both in terms of curricula than topics (e.g. "National Curriculum" > cell, flower).
It's possible to edit items with hundreds of statements with the same property and differentiate with qualifiers, but it's just wont work well with the default interface and can be pratically un-editable. The proposal does have great potential, but it just needs some work.--- Jura 17:32, 4 May 2020 (UTC)[reply]
Support Important to represent and compare key topics covered in different curricula. John Samuel (talk) 18:13, 4 May 2020 (UTC)[reply]
~~Oppose~~ in its current state. --- Jura 18:22, 4 May 2020 (UTC)[reply]
Support per Sadads above. Prburley (talk)
How many statements would you expect for National Curriculum for England (Q17016944)? ChristianKl ❪✉❫ 22:54, 5 May 2020 (UTC)[reply]
- @ChristianKl: I'm not sure exactly, unfortunately the information is in .pdfs rather than a nice table. My guess would be a few 100 although that's only a rough guess, it will be clearer once I start adding it. I'm thinking if there are a huge amount there could be a Wikidata item for each level of curriculum, in this case 'key stages' or some other way of dividing it up. --John Cummings (talk) 23:12, 5 May 2020 (UTC)[reply]
  - @John Cummings: This answer means that you aren't yet clear how to model the domain and thus this proposal is not ready. ChristianKl ❪✉❫ 07:52, 6 May 2020 (UTC)[reply]
    - @ChristianKl: Can you point me to the rule that says you have to have a completed schema before a property is approved? I don't think it is realistic to have a definite schema for this topic before we start exploring the data more given the size of the content (assuming there is at least one curriculum per country), it would require looking through a large chunk of them in detail. The structure of the schema is clear, the only thing that is uncertain is how granular the items are the data is added to. --John Cummings (talk) 08:39, 6 May 2020 (UTC)[reply]
      - We have property proposal discussions to agree on a schema. Without having a specific schema, there's no way to have a consensus for a specific schema. Our policy does say "All opposing points of discussion should be addressed before creation occurs. If there is still disagreement, it is up to the property creator to consider the thoughtfulness and logic behind points of discussion (i.e. an opposing voice with no thought behind it should not block creation, but a single reasonable opposing voice against many supporters may do so)."

As far as I'm concerned the question of having proper examples is still not addressed. ChristianKl ❪✉❫ 08:58, 6 May 2020 (UTC)[reply]

Hi @ChristianKl:

Maybe I can explain the schema here and you can suggest ways to improve explanation above?

What is standardised in this schema: the structure of how information links together, that the property links through from the item about the curricula or a part of the curricula to the subject with qualifiers for start and end dates etc

What is not standardised in this schema: how granular the items for the curricula. This is not possible to standardise because of the range in scale of schemas, e.g the NASA summer school has a very limited number of subjects and activities and so would only need one item, where as the UK National Curriculum has potentially hundreds of them, so would need to be split into more granular parts e.g one Wikidata item for each 'key stage'.

Thanks

--John Cummings (talk) 11:35, 6 May 2020 (UTC)[reply]

You could do an RFC to come up with good samples and once done, update the proposal. There is no point in creating a property when it's not clear how it should be used, especially not to the proposer. Once we have that, it might be easier to form a consensus about the various aspects of the proposal. --- Jura 05:40, 7 May 2020 (UTC)[reply]

Support. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:59, 5 May 2020 (UTC)[reply]
Comment I agree that this will need the curricula to be divided into separate Q-items by stage and probably also by subject, so that each only links to ~<100 items. I suggest this should be done in the proposal examples (also because we don't have a "key stage" qualifier to use the way it is currently written). For comparison, the Schools Online Thesaurus (Q65000093) has 9520 items in its Mix'n'Match, but obviously not all are taught to every stage in every topic. --99of9 (talk) 02:54, 6 May 2020 (UTC)[reply]
Comment I think it's an improvement to replace National Curriculum for England (Q17016944) with Key Stage 3 (Q6398096) in the sample, but a three year period is still too broad. --- Jura 05:40, 7 May 2020 (UTC)[reply]

@Jura1: That is how some curricula are written and genuinely operate (in Australia the stages are 2 years long). But there is another natural break by subject. The reference document is just for Key Stage 3 Science. So we should have an item for that. --99of9 (talk) 05:53, 7 May 2020 (UTC)[reply]

Thanks @99of9: I think this speaks for the need of an improvement in Wikidata to make it easier to have more complex items, lots of the world is complex and can't really be broken down into simpler parts without losing meaning. --John Cummings (talk) 08:46, 7 May 2020 (UTC)[reply]

@99of9: now the property has been marked ready for creation could you press the button to create it please? --John Cummings (talk) 18:14, 7 May 2020 (UTC)[reply]
- There seems to be a disagreement on some of its aspects. Some users supporting the initially proposal, others noting that your sample is still not on spot. --- Jura 04:35, 8 May 2020 (UTC)[reply]

Hi @Jura1: from reading through the comments the only user now opposing the creation of the property is you, is there anything I can do to the example to make it better for you? --John Cummings (talk) 17:29, 9 May 2020 (UTC)[reply]

Some users explicitly support the opposite approach. It seems hard to say they support the current proposal. Personally, I'd be fine with the update suggested by @99of9: and maybe the inclusion of "syllabus" in the description. Somehow I'd hope that you could dig up some more detailed samples at the place you are, but .. --- Jura 18:17, 11 May 2020 (UTC)[reply]

@John Cummings: A proposal marked as ready does not give property creators carte blanche. We have to read the discussion and responsibly take part in it first if anything is unclear or needs to be addressed. As this is quite a big discussion and falls under an area of modelling many probably aren't familiar with, whoever creates it (I would assume some property will be made eventually as there's a demand here) will want to think everything through first. --SilentSpike (talk) 16:34, 11 May 2020 (UTC)[reply]

Thanks for your reply @SilentSpike:, I'm unclear what else I can do, I've answered all the questions and written an entire RFC which outlines future plans. Am I just waiting now for a property creator to review it and make a decision? John Cummings (talk) 20:17, 15 May 2020 (UTC)[reply]

@John Cummings: Yeah you're all good, try not to think of it as anything you need to do. The property creation process can be misleading, it's not a pass/fail for the original proposer, but a collaborative process of refining/challenging the proposal to the point where there's no open questions or issues. So the responsibility for a property creator now (seeing that there appears to be a consensus for creation here) is to read and understand all the points of discussion because we should first participate (i.e. raise any questions/issues of our own) before making the call to go ahead with creation. I personally have not had the time to do so yet as this is quite an in-depth proposal and I'd like to think through the intended modelling and points discussed before feeling comfortable creating this (likely the same for other property creators). --SilentSpike (talk) 20:57, 15 May 2020 (UTC)[reply]

@SilentSpike: thanks very much for the explanation, let me know if any clarifications etc are needed, I appreciate the time spent by whoever reviews it, its gotten pretty long. --John Cummings (talk) 22:02, 15 May 2020 (UTC)[reply]

Comment Maybe it's just an impression of mine, but it seems that some parts of the national curriculum (years < 7) are closer to what I'd call a syllabus (with a definition slightly different from Wikipedia), i.e. half-way between a high level curriculum and a lesson plan.

In terms of what could be mapped to Wikipedia and to specific educational resources, I think this could have more practical value.

The general idea would be to make this accessible to non-specialists (e.g. a parent should be able to a understand a topic from the section of a Wikipedia article and select an educational resource that targets the appropriate level from Wikidata. Maybe this would be as granular as a Khan video. Would this be a good usecase to keep in mind or is it too ambitious? Obviously, if one targets just the level "biology textbook" for "Key Stage 3", one would choose a different approach and probably would even need a new property.

An aspect is whether we should continue to name this "curriculum topics" or if "syllabus" needs be included (personally, I don't think it matters, but maybe the term should be included in the description and the domain clarify that it's covered).

Also, it would be useful to collect actual syllabi (if available/suitable for inclusion): Wikidata:Property proposal/defines curriculum should probably offer "uses syllabus" as an alternate label. :To sum this up, please add the purpose of the mapping in "planned use". --- Jura 18:17, 11 May 2020 (UTC)[reply]

Hi @Jura1:, I've linked from the motivation section to the RFC I created which outlines more of what could be done, you clearly have a lot of ideas around this, please do take part in it. --John Cummings (talk) 09:38, 12 May 2020 (UTC)[reply]

If the purpose of the RFC is to determine the usecases, wouldn't it make sense to first determine these and then formulate a property proposal? Otherwise we just create a vague, possibly redundant property that ends up being deleted later. --- Jura 14:29, 16 May 2020 (UTC)[reply]

Hi @Jura1: I don't understand what more you want me to do, I've written an entire RFC describing the future use cases of the property. --John Cummings (talk) 19:30, 16 May 2020 (UTC)[reply]

Strong Support. Very much needed. Shani Evenstein (talk) 12:56, 26 May 2020 (UTC)[reply]
Weak support Okay, so I've had the time to read through all of this proposal and associated RFC. It's definitely a complex area of modelling and I'm a little on the fence about it. Part of me thinks we should be bold and create this to get started modelling as this gives a good enough starting structure to expose weaknesses in the data and modelling down the line upon which we can build/refine (plus there's a lot of support here). Another part of me thinks this is too general, is it valuable to say "X cirrucula includes the cell". To what level of understanding? Which aspects of the cell are being covered - just the fact that cells exist; their underlying structure; our historical understanding of the cell? It almost seems like we need intermediate items that are instances of such granular educational topics rather than linking directly to the subject items which are instances/classes of things. However, that is problematic because they're not standardised and there are infinitely many ways to slice that cake. --SilentSpike (talk) 13:42, 26 May 2020 (UTC)[reply]