[Data-modeling] Modeling nutritional information

[Jeff Prucher]

>>	2) The name of the topic should ideally be the name of 
>> the product (or a version thereof). E.g., "Kellogg's Corn 
>> Flakes, 18 oz." so that other users can find them and link to 
>> them. (In this case, say if they were trying to model the 
>> product lines of cereal companies.)
> 
> 
>  That makes sense, but I had two worries and I'm not sure if 
> they are ill-founded.
> 
> First, Kellog's works well because the company actually has 
> an apostrophe in the name. But what about a product by 
> Wholefoods? Should it be "Wholefoods' Corn Flakes" or 
> "Wholefoods Corn Flakest"? And, I guess, does it matter?
> 
> Second, is there a need to enforce naming conventions to 
> ensure correct linking? If so, what is the best way to 
> arrange for this?
 
Naming conventions are not technically enforcable, although there are a
number of types that do have naming conventions associated with them. About
all you can do is document the convention in as many places as possible (the
type description itself and property help text for any property that expects
the type).  Possibly, a dedicated user community could keep names somewhat
in line. To answer your first question, it probably doesn't matter, but I
would go without the apostrophe -- it seems to me that most products, when
they are known by brand + name don't use an apostrophe: Apple Macintosh,
Ford Taurus, Kraft Macaroni and Cheese, Canon Sureshot, etc.

		 
> 	3) The name of the "retail data" type should probably 
> be something that more clearly describes the topics that will 
> use that type -- something like "Food product" or "Packaged 
> food" or something like that.  
> 
> 
> Agreed. But also, would that include drinks like milk, or 
> should a different type be created? 
 
I would include drinks and eats in the same type for this, since
nutritionally, it doesn't matter whether it's liquid or solid. So maybe
"food" will be misleading for some people. It could be simply "Food or Drink
Product" or the like (and doesn't that phrase make it sound tasty?)
 		 
> 	4) I think it would be great to find a way to make the 
> nutrient information more flexible, rather than only 
> providing a handful of properties. Different countries are 
> going to require different things on the labels, and it would 
> be nice not to have to change the model for every market.  I 
> don't have a good idea just off the top of my head, but it's 
> worth thinking about.
> 
> 
> What I have so far includes all the labels I could find on UK 
> and US products, apart from RDA. I do not like the RDA 
> because it can change, whereas the absolute values are 
> constant across time. In other words, if eventually somebody 
> in the USDA does realize that the RDA should be reduced given 
> the levels of obesity, or even worse, that the RDA should be 
> tailored to the individual's exercise levels, all old RDA 
> data would be invalid and not available for comparison. But 
> using the absolute values would avoid this problem.
> 
> Does what I am saying make sense? At any rate, all I am 
> declaring is that I would never use the RDA field, not that I 
> would block others from using it. 
 
The USDA has a lovely, copyright-free database with quantitative information
about packaged food, which measures an insane amount of nutrients. You can
search the data here: http://www.nal.usda.gov/fnic/foodcomp/search/ and
download it here: http://www.ars.usda.gov/Services/docs.htm?docid=8964
 		 
Jeff


> 	 
> 	 
> ________________________________
> 
> 	From: data-modeling-bounces at freebase.com 
> [mailto:data-modeling-bounces at freebase.com] On Behalf Of 
> Mohammad Al-Ubaydli
> 	Sent: Sunday, August 24, 2008 12:23 PM
> 	To: data-modeling at freebase.com
> 	Subject: [Data-modeling] Modeling nutritional information
> 	
> 	
> 
> 		Hi all,
> 		I am starting a model of nutritional 
> information on food labels and can tell that I am doing it 
> wrong even after reading the documentation... Could someone 
> help me with this model?
> 		
> 		
> http://www.freebase.com/view/user/idiopathic/nutritional_information
> 		
> 		What I want is for someone to easily type in 
> the barcode of a food item and get back the full nutritional 
> information. So I made two types, one called "Retail data", 
> and the other "Nutritients". The "Nutrients" type includes 
> the "Retail data" one.
> 		
> 		However, when I try to add a record, it insists 
> on a name. I tried this one:
> 		
> 		
> http://www.freebase.com/view/guid/9202a8c04000641f8000000008ff
> a1d8?domain=%2Fuser%2Fidiopathic%2Fnutritional_information
> 		
> 		Apart from telling you all what I had for lunch 
> today, this does not seem to me like a sustainable naming 
> strategy. The primary key of each entry should be its 
> barcode, not the compound name of the product and retailer.
> 		
> 		What am I doing wrong, and how should I fix it? 
> (If anyone would like to help me fix this, I would be 
> delighted to invite you to administer the domain with me of course.)
> 		
> 		Many thanks for any help, and my apologies in 
> advance for being such a newbie.
> 		mohammad
> 		
> 		Dr. Mohammad Al-Ubaydli
> 		e me at mo.md
> 		w www.mo.md
> 		
> 
> 
> 	_______________________________________________
> 	Data-modeling mailing list
> 	Data-modeling at freebase.com
> 	http://lists.freebase.com/mailman/listinfo/data-modeling
> 	
> 	
> 
> 
>