Wikidata:Property proposal/food composition
Jump to navigation
Jump to search
food composition[edit]
Originally proposed at Wikidata:Property proposal/Natural science
Withdrawn
Description | nutritionally important components of foods |
---|---|
Represents | food composition (Q5465452) |
Data type | Item |
Template parameter | "fat", "satfat", "transfat", "carbs", "fiber", "protein", etc. in en:template:nutritional value |
Domain | food product (Q951964), recipe (Q219239) |
Allowed values | fat (Q127980), protein (Q8054), carbohydrate (Q11358), iron (Q677), etc. |
Example 1 | Big Mac (Q506796) → fat (Q127980) quantity (P1114) 28 gram (Q41803) |
Example 2 | Big Mac (Q506796) → protein (Q8054) quantity (P1114) 25 gram (Q41803) |
Example 3 | Whopper (Q1135327) → carbohydrate (Q11358) quantity (P1114) 49 gram (Q41803) |
Example 4 | Baconator (Q4839954) → iron (Q677) quantity (P1114) 5.25 milligram (Q3241121) |
Source | https://fdc.nal.usda.gov/ |
Planned use | Move data from infoboxes into Wikidata |
Expected completeness | always incomplete (Q21873886) |
See also | has listed ingredient (P4543), has active ingredient (P3781) |
Motivation[edit]
I was surprised this property is missing. This seems neccessary to include data about the food composition (Q5465452) of an item. This property should also be distinct (i.e. an item cannot have fat (Q127980) quantity (P1114) 28 gram (Q41803) and fat (Q127980) quantity (P1114) 34 gram (Q41803) at the same time). U+1F360 (talk) 15:40, 29 October 2019 (UTC)
Discussion[edit]
Notified participants of WikiProject Food
- Support David (talk) 13:13, 30 October 2019 (UTC)
- Comment Would has part(s) of the class (P2670) work in this situation with mass (P2067) as a mandatory qualifier? Dhx1 (talk) 13:16, 30 October 2019 (UTC)
- @Dhx1: I don't think so because has part(s) of the class (P2670) has a data type of
item
where what is needed is to specify thequantity
. U+1F360 (talk) 14:19, 30 October 2019 (UTC)- I agree with Dhx1. We could do Baconator (Q4839954)has part(s) of the class (P2670)iron (Q677)
quantity (P1114)5.25 milligram (Q3241121) (with quantity (P1114) instead of mass (P2067)). --Tinker Bell ★ ♥ 22:55, 1 November 2019 (UTC) - @Tinker Bell: Why has part(s) of the class (P2670) and not has part(s) (P527)? (Perhaps I don't understand the difference between the two). U+1F360 (talk) 12:42, 19 May 2020 (UTC)
- @U+1F360: I think has part(s) (P527) is useful for listing all the parts, that together, conforms a unique item or collection, like a single band (e.g The Beatles (Q1299) has part(s) (P527) John Lennon (Q1203), Paul McCartney (Q2599), etc,), a confederation (United States–Mexico–Canada Agreement (Q56839716) has part(s) (P527) United States of America (Q30), Mexico (Q96), Canada (Q16)); while has part(s) of the class (P2670) is used when the component parts are general subclasses. In Spanish the names are clearer, and has part(s) (P527) is called 'compuesto de' ('composed of') instead of 'has part'. --Tinker Bell ★ ♥ 05:41, 21 May 2020 (UTC)
- @Tinker Bell: Why has part(s) of the class (P2670) and not has part(s) (P527)? (Perhaps I don't understand the difference between the two). U+1F360 (talk) 12:42, 19 May 2020 (UTC)
- I agree with Dhx1. We could do Baconator (Q4839954)has part(s) of the class (P2670)iron (Q677)
- @Dhx1: I don't think so because has part(s) of the class (P2670) has a data type of
- of (P642) is a property that translates poorly and I Oppose the model that uses it. Tinker Bells solution seems better but I'm open to finding a even better model. ChristianKl ❪✉❫ 22:11, 4 November 2019 (UTC)
- @ChristianKl: I removed of (P642) from the model as requested. U+1F360 (talk) 20:43, 2 January 2021 (UTC)
- Comment How will this work for foods/products that have different compositions in different countries, etc? Seems like it could become an unwieldy mess of qualified statements. I think a better approach would be to add external identifiers for the various food composition databases. NoInkling (talk) 22:22, 5 November 2019 (UTC)
- I would assume a new item would be created for each "package" (or each SKU, etc.) which might happen anyways... U+1F360 (talk) 22:26, 5 November 2019 (UTC)
- Comment I would really like to see some way of capturing nutritional information, like energy content, carbohydrates, protein, fiber, per mass/volume. Iwan.Aucamp (talk) 15:41, 17 May 2020 (UTC)
- @Iwan.Aucamp: See food energy (P7971) for energy content. U+1F360 (talk) 18:08, 17 May 2020 (UTC)
- Nutrients (carbs, iron, vitamin b12) and ingredients (water, beans) should have two different properties Nepalicoi (talk) 18:28, 19 May 2020 (UTC)
- @Nepalicoi: We could use object has role (P3831) as a qualifier to differentiate the two. Perhaps ingredient (Q10675206) and food composition (Q5465452)? U+1F360 (talk) 19:05, 19 May 2020 (UTC)
- or perhapse nutrition information (Q2006410) is more appropriate? U+1F360 (talk) 19:31, 19 May 2020 (UTC)
- Support for using it for food composition (synonymous for nutritional information), using datatype item using proportion qualifier. Ingredients should be taken to has listed ingredient (P4543), which already exists. Germartin1 (talk) 18:25, 6 June 2020 (UTC)
- Oppose The model proposed by User:Dhx1 and User:Tinker Bell is more intuitive to me. --99of9 (talk) 08:02, 22 July 2020 (UTC)
- I have updated the proposal to be more consistent with has listed ingredient (P4543) and has active ingredient (P3781). I think being a distinct property is valuable rather than overloading has part(s) of the class (P2670). I can open a new propsal if we want since this is different than the initial proposal. Overall I think this is better, appreciate the feedback! U+1F360 (talk) 21:45, 17 October 2020 (UTC)
- Are there any more objections or can this be created? U+1F360 (talk) 00:31, 28 December 2020 (UTC)
- Comment Another way to approach this would be to create individual properties for the nutrients themselves with the data type quantity. These properties would then be available for use across food items. The advantage of this approach is that each statement could be referenced to the source where the info is found. If we follow the qualifier strategy then it will be more difficult to indicate references. Food composition values are different depending on the food composition table or database from which they are sourced. The way the current proposal is framed, it seems like this would only accommodate food composition data from the USDA's Food Data Central. YULdigitalpreservation (talk) 14:03, 31 December 2020 (UTC)
- @YULdigitalpreservation: I'm happy with seperate properties, but it felt like a lot of properties. If we'd rather create properties for each type that makes more sense to me in a lot of ways. There could be some weird edge cases though. For instance, if we created a property for carbohydrate (Q11358), would we create another property for sugar (Q11002) or would it use a qualifier on the carbohydrate (Q11358) property? U+1F360 (talk) 18:01, 31 December 2020 (UTC)
- @YULdigitalpreservation: I created Wikidata:Property proposal/carbohydrates as an alternative to this proposal. I would appreciate your feedback. :) U+1F360 (talk) 17:28, 2 January 2021 (UTC)
- Comment I concede that has part(s) of the class (P2670) solves the need as can be seen by the example in Chef John's Buttermilk Biscuits (Q95382239). Please close this proposal. Thanks! U+1F360 (talk) 22:32, 2 January 2021 (UTC)
- Withdrawn by proposer. --Tinker Bell ★ ♥ 02:10, 21 January 2021 (UTC)