Translating between VF and DataFood?

[bhaugen]

We're talking about doing an experiment here: https://www.loomio.org/d/lfpOvdhB/data-food-consortium%3Ffrom=1

We want to try to translate between DF to VF and back, to see what problems we run into.

Do you have any examples of use of the DataFood Ontology? Especially the concepts in Product Reference and Real Product boxes.

I think they would help us understand better as well as provide test cases for the experiment.

[bhaugen]

@myriamboure ? @djodjoni ?

Also, have you or anybody associated talked about ontology for recipes?

[djodjoni]

I am curious of the output of this exercise as I would use those ontologies together, rather than just map them, similarly to the one that I use https://w3id.org/openfooddata/onto/core as I see them serve a little different purposes :)

btw @myriamboure I am curious are there some news on this ontology and your cooperation with the other french companies including thefoodassembly ?

for recipes http://schema.org/Recipe is quite OK for me :)

[bhaugen]

@djodjoni thanks a lot. Will study http://schema.org/Recipe . From a quick glance, I don't see how to handle intermediate resources that are outputs from one step and inputs to the next step: like dough gets made into loaves, or even more complex, dough goes thru several steps itself (autolyse, raise, knead, raise again, knead again, etc). But maybe I missed something, it's got a lot of moving parts.

[myriamboure]

Hi there ! @bhaugen I'm super busy in Australia with the whole OFN team at the moment. We didn't update already the owl file but here is the image of the latest version of the Data Food Consortium business ontology, it has changed quite a lot. The recipe in my mind is the "composition" node (the ingredients and proportions), and then we have a "transformation" node (which can alter both the nature and location of the product), and for this one you have a planned one (the recipe) and the realized one (the actual implementation of the recipe). It is not yet completely stable I think, needs a few more iteration, but I'll be super happy then to have a call with you and see how we connect with Value Flow and schema.org, and converge. We don't want to maintain ourselves an ontology, but this exercise is super usefull and can also make general ontologies evolve to better meet our needs.

[djodjoni]

@bhaugen in just the context of a general food recipe you can use http://schema.org/recipeInstructions, but I wouldn't use this in a general VF in-out chain of processes

for transformation-like recipe in order to be tracked and part of value flow I don't have a match but VF, and I don't need another one and this is what I mean Vocabs can be combined rather just mapped.

[bhaugen]

@djodjoni thanks again.

@bhaugen in just the context of a general food recipe you can use http://schema.org/recipeInstructions, but I wouldn't use this in a general VF in-out chain of processes

Yeah, we really need the input-process-output flows, especially for food processing. Think base ingredient that then gets used in lots of other recipes, in addition to the use cases I mentioned upthread.

In manufacturing systems, those kinds of structures originated in food processing software in the 1990's. But they work for anything.

for transformation-like recipe in order to be tracked and part of value flow I don't have a match but VF, and I don't need another one and this is what I mean Vocabs can be combined rather just mapped.

We agree, and as you know, are doing that in some cases. The reason we want to translate to and from the DataFood vocab is that we have similar but not exactly equal concepts. If I understand correctly, the DataFood gang looked at VF and thought it was too abstract for the food domain. Which is true. VF needs to be very general-purpose, DF needs to be about food. So what if some group is producing food and some other group (I'm looking at you, Sensorica) is developing composting equipment? How do they combine in an economic ecosystem? I mean, when their operational systems need to talk to each other?

(Plus, we have developed specialized systems for food ourselves, and thought the translation from the general to the specialized and back was coming up...:-)

[myriamboure]

Hey @bhaugen we have quite moved forward on lots of iteration of the DFC ontology, have a look at the last blogpost http://datafoodconsortium.org/blog/business-ontology-and-product-thesaurus-v-1-published-after-a-year-of-regular-iterations, we just published a new version of where we are at on the repo :-) Happy to discuss with you, might be interesting to share a bit on where we ended up with and se if there would be some more interesting connexions to establish. Cheers!

[bhaugen]

Hey @myriamboure thanks for the news. Unfortunately, I am swamped with work until next week, when I will get deep into this. I am very interested.

[fosterlynn]

@myriamboure @kirstenalarsen Rachel

I said I would start mapping the DFC ontology with VF. I did an initial conceptual mapping with the help of @elf-pavlik and @bhaugen . It is here: https://docs.google.com/document/d/1ODah5xKw_QkVOLbJlKcebCfHe6LUG08Ph1X35Y_Sn0o/edit?usp=sharing.

Comments welcome! And maybe in January we can figure out where to go next... ?

[fosterlynn]

@myriamboure hello again!
cc @elf-pavlik @bhaugen

We are looking at your ProductBatch because we would like to break out our lot identification in that way, it is better than what we have now, and I expect food has the most stringent lot/batch requirements. The reference I am using is this https://github.com/datafoodconsortium/ontology/blob/master/DFC-%20Concepts%20definitions.pdf.

Are ProductBatch and its properties fairly stable and are there any issues we should be aware of if we decide to reference it directly from VF? (Still under discussion, just wanted to check in here as part of our decision process.)

And thanks!

[myriamboure]

Hello @fosterlynn !

I feel so sorry, pretty embarrased to not have answered before, you know in he rush of priorities and emergencies some things sometimes are moved from weeks to weeks to weeks... so sorry ! But please in that case feel free to ping me again so it goes up on the emergency list :-)

I'll review now your mapping doc, but to answer first your last question, what I would say is that conceptually it is pretty stable, BUT we have not yet used it in a real life case so in that sense for me, until it is used in a real life case and we have proven it does the job, it is not stable... I know this answer is not helping :-o I don't think it's the most risky concept if that can reassure you ;-) It's a pretty basic one, a "product batch" is defined by an id, a number basically. And if you change something on your machine, or on the input products, or on the date of picking or slaughtering, it's a new batch. Maybe the written definition should actually slightly evolve to include more than production date and peremption date, we should say more broadly that it is a set of physical product that have the same characteristics, have been through the same process, at the same time, and can be consider as "of the same inner consistency and quality"... something like that. Do you agree ? I'll ask Bernard our ontology what he thinks about it ;)

I'll do the mapping review now !

[myriamboure]

Also for info @bhaugen @fosterlynn @elf-pavlik we need to fine tune a detail about the LDP standard but we are about to publicly release the first iteration of the technical standard of DFC : https://datafoodconsortium.gitbook.io/dfc-standard-documentation/

[fosterlynn]

Also for info @bhaugen @fosterlynn @elf-pavlik we need to fine tune a detail about the LDP standard but we are about to publicly release the first iteration of the technical standard of DFC : https://datafoodconsortium.gitbook.io/dfc-standard-documentation/

Heartiest congratulations!!! :heart:

[myriamboure]

Thank you @fosterlynn ! I reviewed the document and made some comments. I would need to deep dive into Value Flow ontology to totaly enter the logic but I have the impression we came up with pretty close solutions actually on lots of aspects ! In some you were more generic and didn't cover, in other we went a bit further, and the way we describe products is probably to complementary way, but different to describe the same reality. I think if/when we want to move forward it might be interesting to plan some real time working session to see if there are parts where it would make sense to converge or if it's just good to keep the two logics and connect both ontologies. I wouldn't have time to be honest in the coming months as we are 200% on the creation of the local OFN coop in France and I'm going to have a baby soon (!) but I like the idea to try to go one step further ! But real time conversation will be required... my priority for now on DFC is to make sure we build an operational prototype to have real life cases to stabilize our ontology and check if we modelized well... and have something to show to find money to make it easier to move forward (as we need to pay the beer ;-)) If you really think we should do it sooner we can try to find a 1/2 work during the summer in our time compatibility zone ;-) Cheers !

[djodjoni]

Hi all :) @myriamboure congrats for the baby !!! i've been away for a loooot of time and will still be.. :) I also got a second boy some 9 months ago :) and been even more terribly busy since :) I've been away also from my food project for some time but just saw some activity and decided to share a thought :) I was building a prototype mostly based (also mixed with valueflows) on GS1 ontology (https://www.gs1.org/voc/) which itself is based on goodrelations and apart for the valueflows part it provides quite broad coverage on products, packaging, pricing, delivery etc. quite complete IMO and was able to cover most of the cases for short spply that I've encountered . On top of that i've noticed they are also working towards global identifiers URIs https://2019.semantics.cc/robust-identifiers-leading-quality-data https://www.gs1.org/standards/Digital-Link/1-0

I know this is meant for big brands etc, but the approach and directions seems quite good and @philarcher1 seems to be quite into semantics and the whole idea about the 'good' www :)

So I just thought it could be interesting to check if you can get some useful stuff out of it or probably reuse part of it :)

After all the biggest issue of semantic web by far still remains the adoptions and agreements over ontologies :)

Cheers :))

[fosterlynn]

Babies!! :baby: My "babies" are 30 and 33 years old, but I definitely remember how busy it was. And wonderful too. But really really busy!

@myriamboure I will read and reply to your comments, thank you for doing that! But just to say in general, I agree, let's let it sit for now, there is no rush. I'm guessing that we will have good practical reasons to come together at some point, and then we can spend the time for real working sessions. Maybe around a specific project so that we have a focus? One that might come up, but not very soon: We have a project going in holochain called HoloREA, which will implement ValueFlows an an economic framework for other projects to build upon. There are several food-related projects who want to use it. As a small connection, they have been talking together about how to identify their products/resources using taxonomies, as they form food networks themselves and with each other. I sent your doc about using facets along with taxonomy, I think that would help them a lot to think about it. Also as we work with them more, we will see what is required for more specific food ontology or if VF covers their needs.

@djodjoni hey, good to e-see you! Thanks for the links, I'm particularly interested in the global identifier question, but will look at all of them. I was noticing the other day the amount of detail that Good Relations provides on the whole e-commerce sales aspect. If you get back to your prototype, I'd be interested in seeing it!

[fosterlynn]

@myriamboure I will read and reply to your comments

Done. Don't worry about continuing the discussion until it becomes more of a priority. But if you do, then I'll pay attention. If not, I won't worry.