Talk:Mika/Temp/WikiFCD: Difference between revisions

From WikiDotMako
Line 221: Line 221:
| common name (P76) || [http://wikifcd.wiki.opencura.com/prop/P76] || common name (P76)  
| common name (P76) || [http://wikifcd.wiki.opencura.com/prop/P76] || common name (P76)  
|-
|-
| SMILING Food composition table for Vietnam food code (P77)|| [http://wikifcd.wiki.opencura.com/prop/P77] || SMILING Food composition table for Indonesia food code (P78)
| SMILING Vietnam food code (P77)|| [http://wikifcd.wiki.opencura.com/prop/P77] || SMILING Indonesia food code (P78)
|}
|}



Revision as of 17:58, 27 October 2020

Related Work

This paper from ISWC 2019 describes a knowledge graph that includes nutrient data. [1] Data: [2]

Wikiprojects to notify

What's the best to notify them? Mika (talk) 12:33, 20 January 2020 (EST)
I think there is a ping project template that we can use. Ping Project. Hweyl (talk) 19:51, 21 January 2020 (CET)
Great! Is the plan to say that we'll build this in a way that'd be easy to be incorporated into WikiData if they choose to do so in the future? Mika (talk) 22:17, 21 January 2020 (CET)
Draft text: Greetings members of WikiProject Food! We'd like to let you know about a complementary set of work we are planing related to food composition data. We are planing to create a Wikibase just for food composition data. To support this work we are proposing a Grant. We invite your review of the description of the project so we can learn of any feedback you may be willing to share with us.

People to review

Gene Wiki

About WD

  • draft response
  • Thank you for the feedback! We have been thinking about this for a while as we'd initially planned to do this in Wikidata. Knowing how variable the types and depths of information are in FCDs, it is possible that we will include data that may never be appropriate for Wikidata. We decided that it'd make sense to build a Wikibase where we can hold all the details.

Example datasets may look like this FAO data on detailed information on phytate or more standard data which includes aggregate phytate data. And this is just one nutrient; there are many more nutrient data as well as meta data. We believe that it is important have these discussions in WD. We also feel that the discussion on WD and this Wikibase database could develop simultaneously.

Our approach includes the creation of ShEx schemas Wikidata:WikiProject Schemas we will publish these schemas in Wikidata's E namespace for entity schemas. This way the data models that we are discussing could also be shared and discussed on Wikidata. Our approach is to prepare data in a Wikibase and then help coordinate getting the data into Wikidata once the Wikidata community has built consensus on how much nutrient data will be appropriate for Wikidata.

Response to other project

Happy to clarify these points!
First, we want to emphasize that our focus in this project is the re-organization and standardization of the existing databases and we will do our best to classify and store every bit of information from each database. We will not limit the amount of data to be included in our Wikibase instance, with the hope that different communities, such as Wikidata, can pick and choose what to include in their own databases. Our Wikibase instance can serve as a place to sort data from all the databases that are available on the Internet, including ones you mentioned, into one place, so that users can pull, combine, and analyze necessary data more easily. We believe that this project will be able to offer something these FCDs currently do not/cannot do, by harnessing the power of peer production.
I use many of these FCDs as a nutritional epidemiologist for research, and also as a migrant individual who records dietary information related to foods from my native country as well as other countries. The current situation of having multiple incomplete databases in various formats is much less than ideal for meeting the needs of diverse communities and individuals. For instance:
* I keep my daily dietary records in English. The software I'm using primarily uses the data from USDA.
* I sometimes eat foods more commonly consumed in Japan like 海ぶどう.
* I look for this item in the USDA databases, using several keywords including its scientific name, but I can't find the information. Perhaps this data exists in another database but I'd need to check each database one by one.
* I use another algae item as a substitute in my record but the nutrient data are available in the Japanese database.
This is just one example of the kinds of problems people may run into because of the incomplete connectivity among the existing databases. A global FCD can open up ways to explore many, many more new questions and solutions not just in food and nutrition but in many other topics that Wikimedians may be interested in. We believe that having this placeholder for all FCD information is a good way to contribute to different Wikimedia projects.
I believe OFF and our Wikibase instance take distinct approaches. According to their website:
"Open Food Facts is a database of food products with ingredients, allergens, nutrition facts and all the tidbits of information we can find on product labels."
OFF builds up the database by individually contributing nutrient data from food product labels. Our project's approach is different. We will be using existing databases and compile them into a standardized and structured database. As I mentioned before, OFF and our Wikibase instance are complimentary. On of the strengths of OFF is the ability to have product nutrient data that are not yet in larger public databases like USDA databases. Combining OFF and our Wikibase instance, we can have a more comprehensive FCD than any single database.
Great point about the plan beyond data importation. Like any Wikimedia project, peer production has the potential to actually keeping information more up-to-date than any working groups with limited numbers of participants. Any methods we employ to import, check updates, and maintain this database will be documented, so that future participants can easily learn how to do each of these activities and start contributing to the project. We will engage in outreach activities to involve diverse participants and we hope that documentation/tutorials and community engagement will increase the chance of frequent updates of the data.
In short, there is an enormous amount of food composition data on the Internet already. There have been several attempts to combine some of these databases but none has succeeded in creating a comprehensive and easy-to-use global database. Open and collaborative peer production has its advantage over smaller working groups in compiling these databases into one place and maintain the data. We believe that this is a powerful solution to the decades-old problem in nutrition and will benefit a wide range of audiences, from Wikidata users to FCD developers.
We hope this clarifies your questions. Thanks for engaging in this topic!

Response to alex

Thank you so much for the comments and suggestions! I love the idea of digital community engagement through content drive/contest. We will work with the community engagement intern to plan and incorporate this idea. We will edit the proposal to add this point.
We hope that our wikibase will be useful to OFF and we'll be able to demonstrate seamless data exchanges between the two. I completely agree that this is an important knowledge base for several SDGs and we look forward to working with diverse communities to improve it.
Thanks again for the feedback!

Wikibase for prototyping wikiFCD

Testing Wikibase

  • Main page [4]
  • List of properties [5]

Inventory of FCTs

  • We may want to consider using our wikibase to inventory the FCTs of interest.
  • example item for FCT [6]
YES!!! Mika (talk) 22:18, 12 March 2020 (CET)
Great! Thank you. As I work through the list on FAO/INFOODS, I'm discovering more and more about the complexity of these databases. Regional databases are tricky as they combine existing databases from different countries and sometimes add new information. The easiest thing to do may be to stick with single-country databases that are not based off another database...will keep you posted. Mika (talk) 03:12, 21 April 2020 (CEST)

Useful links

response to the committee

Thank you so much for reviewing our proposal! We wanted to respond to the three major lines of criticism raised in our proposal.

1. Relationships with existing initiatives

The most important issue raised by reviewers were concerns that our project overlapped with existing initiatives, Open Food Facts (OFF) in particular. This issue was raised on the talk page for our proposal during the discussion face but reviewers felt that our response there was not convincing.

We have made major changes to the text our proposal to try to explain in more detail how this project and OFF are complementary and to try to explain the difference between nutrition-label data (OFF's domain) and food composition data (WikiFCD's). Although they are related, they are different data with different sources, different audiences, different challenges, and different uses. Food composition dat is best understood as a "downstream" source of granular data for projects like OFF as well for Wikibase instances like Wikidata.

We have spoken in depth with Stéphane Gigandet (the founder and leader of OFF) who has in turn spoken about our project with the OFF board. Gigandet is excited about our project and, with support of the OFF board, and has graciously added his name to the list of advisors for our proposal. Part of the reason Gigandet is excited about proposal is that OFF has previously attempted to incorporate some fruit and vegetable FCD from USDA and CIQUAL. After running into some of the issues our team discusses in this proposal, like extremely granular data and divergent and shifting formats, OFF decided not to move forward with supporting the types of FCD our proposal targets. With Gigandet's advice, we will work closely with OFF to ensure that WikiFCD not only does not compete or duplicate effort with OFF but that it provides a useful resource that they can draw from.

Although it was not raised in the reviews, a new Wikibase instance like WikiFCD will reduce the burden for other communities such as Wikidata as well so that they can focus on their main project aims. We have updated our proposal to make it clear that we will work closely with Gigandet to integrate our work with OFF as a way of demonstrating how other Wikibase instances can incorporate data from WikiFCDs.

2. Community engagement

A number of reviewers raised concerns about our ability to build an engage a community. Although we agree that this reflects the biggest challenge and risk for this project, we believe that, with a WMF grant, we will have the resources we need to succeed.

If our project has less in the way of existing community support than some other grant proposals, it is because our goal is to engage new groups of experts in the WMF ecosystem rather than calling upon already overtaxed Wikimedia volunteers. The type of outreach we are proposing will involvement building partnerships which could lead to expansion and diversification of Wikimedia contributors. Our approach is to provide domain experts experiences with Wikimedia systems and tooling that they find valuable. This strategy for engaging domain experts is consistent with the findings and recommendations of the GeneWiki program of work.

User:Hackfish is an established academic expert in global health and nutrition. She is currently working at both San Francisco State University and Harvard School of Public Health and will be starting as an Assistant Professor at the Johns Hopkins University in September 2020. User:Hackfish is well positioned to user her deep connections in the academic nutrition community to help this project succeed and this engagement project will be the large part of what the intern will work on. We have already garnered strong interest in this project among the nutritionists at both Harvard and Johns Hopkins and will be working with teams at both places to contribute to WikiFCD and to engage broader communities.

3. Budget Question

There was a lack of clarity in our proposal about what the intern would be doing for 8 hours each week. We have edited the proposal to clarify that the intern will be working with the project manager to create online learning tools and seminars to develop and deliver curriculum focused on teaching nutrition professionals to use and contribute to WikiFCD as well as about peer production, WM projects, and ways to contribute to Wiki-based projects in general. We aim to employ a student who is interested in working with online communities who can support us one day a week for two semesters so as not to create burden on their workload and interfere with their academic work. We are confident that we will be able to identify such an individual.



Integrate into the proposal:

Based on the in-depth discussion both with various WM members and nutrition researchers over the past year, we decided that it would be most useful to diverse communities with diverse needs if we build a comprehensive Wikibase instance for all existing FCDs from around the world. While some communities may be only interested in the most common nutrient information (e.g. total fat, trans fat, calories, sugar), other communities may want more specific information (e.g. Phytic acid (by HPLC/HPAE) : Zinc ratio) or other information related to the food item (e.g. scientific names, varieties of fruits, geo-locations).

WikiFCD can make significant contributions to various WM communities with interests in nutritional data. The Wikibase we are creating will be an expert-curated data set that is mapped to Wikidata. The Wikidata community as well as any other wikibase community will be able to reuse entity schemas or entity data (or both!) from our system. Data will flow back into Wikidata if desired. We strongly believe that WikiFCD will make a positive impact on WM and other communities by accommodating their varying needs and bringing more equitable access to an easily usable database. This Wikibase instance is a new and innovative approach that addresses problems that need to be, and yet have not been, solved.




Comments from reviewers

1. Relationships with existing initiatives

  • I am encouraged to see this project proposal suggesting looking into another aspect of knowledge gaps. I am concerned there are already existing initiatives and wondering why the proposers have chosen to not to align with those.
  • It does fit with Wikimedia's strategic priorities. However, it looks like a competitor to "Open Food Facts", another free software and open data but the concept of both projects is different.
  • I do not find this project to be critical to the current state of knowledge. There are existing resources they could join to do this work.
  • It does fit with Wikimedia's strategic priorities and the budget is reasonable but not enough community engagement. There is a need for a wider or major community discussion to ascertain whether this could be hosted on Wikidata or not. However, it looks like a competitor to "Open Food Facts", another free software and open data but the concept of both projects is different (from the proposal & answers to questions on the proposal talk page).
  • A lot of concerns about the creation of another instance. The answers in the discussion page don't convince me.

2. Community engagement

  • A new instance of Wikibase? Without a community?
  • It does fit with Wikimedia's strategic priorities and the budget is reasonable but not enough community engagement. There is a need for a wider or major community discussion to ascertain whether this could be hosted on Wikidata or not. However, it looks like a competitor to "Open Food
  • Interesting concept but some concerns about the methodology.
  • They really need to become more involved since they were not able to get any endorsements.
  • The proposal has very little community engagement with current Wikipedia communities.
  • This seems iterative, but minimally so.

Grant proposal edits

Through this pilot project, we will write schemas to describe our data model based on five large food composition datasets that are already available online and develop good documentation for both project development and use. The focus on equity and global nature of the project requires diverse participants, which is very much in line with the missions of projects supported by the Wikimedia Foundation. Through this pilot project, we hope to show how peer production can contribute to the improvement in data/knowledge disparities in global nutrition. We believe that WikiData is an awesome way to build connections between a range of free culture related nutrition projects like Open Food Facts that might do the same.

We will test several automated and manual methods to populate the wikibase with nutrient data from 5 food composition databases from around the world (see the Project Plan section section for details). We will write schemas to describe our data model. We will map our properties to Wikidata properties.

2. Why is this a good idea?

  • First, this Wikibase instance will significantly improve the usability of FCD from different sources for diverse users - from WikiProjects and Wikipedia editors and viewers to academic researchers to public health workers. WikiProject food and Drink on English Wikipedia and its equivalents in other languages are universally popular WikiProjects among editors and likewise, many articles on food and drink are within the top 10% of any Wikipedia's articles by pageviews. This new project can contribute to a topic that is of high interest to many people.
Building a structured dataset is also a key step in identifying most appropriate data to borrow in resource-poor settings where up-to-date, detailed, and regionally appropriate FCD are not readily available. This new database will also open up ways to explore new research questions to explore more nuanced nutrition data (e.g. changes in nutrient content of the same product, depending on the climate conditions of the year), which can potentially make substantial advances in nutrition and health research.
  • Secondly, by creating an instance of Wikibase for this project, we will be able to design our own data models, with input from Wikidata, to incorporate data from heterogeneous data sources. If subsets of the data are appropriate for Wikidata, we will be able to provide machine-actionable ShEx schemas that will help us prepare data for other systems. In this way the data will be readily-available for incorporation into Wikidata if desired.
UPDATE
In our recent communication with the Open Food Facts community, they discussed that OFF is in fact interested in using data from USDA (USA) and CIQUAL (France). However, it is burdensome having to deal with diverse and dynamic formats - they mentioned three separate format changes in USDA since they started looking at this. Having another project like WikiFCD can help each community focus on their main project goals instead of each having to deal with these issues we raised in this proposal. This conversation with OFF reinforces our belief that WikiFCD will be helpful to diverse peer production communities.
  • Finally, we will complete this project with diverse communities from around the world as these FCD can be translated into/from many languages. The design of Wikibase will allow us to more easily support additional languages in the data itself, as well as in user interfaces.

Brown Bag Seminar

Some ideas for things to present:

  • slides on the project idea, our vision for the overall process
  • slides on the process of "automated" reading of databases - maybe USDA API example or SMILING excel sheet example
  • slides on connecting to other Wikibase instances - wikidata example
  • demo on how a contributor can easily identify which language is missing the food item in wikidata, which makes it easier for contributors to see where they can make contributions in terms of translation. (I forgot the name of this tool...)
  • demo on making a query based on research questions (e.g. nutrient content differences across databases; comparison of numbers of nutrients included in different databases; Wikipedia articles mentioning a particular food item etc etc)
  • demo on how to convert data to csv or other formats that can be used in R, STATA etc.
  • slides on our plans for the community expansion and engagement (what we will do initially ourselves; what peer contributors will do once the system is ready)

Kat flag

  • Bots: We wrote a series of bots using the WikidataIntegrator python module [7]. These bots can be used to read in data from a source and then create statements in the Wikibase according to our data models. As of November, 2020 we have written bots to:
  1. add countries to the Wikibase (sourced from Wikidata)
  2. add taxon names that have a GRIN id (sourced from Wikidata)
  3. add human languages (sourced from Wikidata)
  4. add USDA Food Data Central (sourced from FDC's API)


Properties used by FCT

😊 Vietnam link 😊 Indonesia FDC link
water (P5) [8] water (P5) water (P5) [9]
energy (P6) [10] energy (P6) energy (P6) [11]
Protein (P7) [12] Protein (P7) Protein (P7) [13]
total lipid (P8) [14] total lipid (P8) total lipid (P8) [15]
Ash (P9) [16] Ash (P9) Ash (P9) [17]
carbohydrate (P10) [18] carbohydrate (P10)
dietary fiber (P11) [19] dietary fiber (P11)
Calcium (P13) [20] Calcium (P13)
Iron (P14) [21] Iron (P14)
Zinc (P19) [22] Zinc (P19)
Vitamin C (P24) [23] Vitamin C (P24)
Thiamin (P25) [24] Thiamin (P25)
Riboflavin (P26) [25] Riboflavin (P26)
Niacin (P27) [26] Niacin (P27)
Vitamin B-6 (P29) [27] Vitamin B-6 (P29)
Folate, total (P39) [28] Folate, total (P39)
Folate, DFE (P42) [29] Folate, DFE (P42)
Vitamin A, RAE (P45) [30] Vitamin A, RAE (P45)
Vitamin A, IU (P49) [31] Carotene, beta (P46)
Retinol (P75) [32] Retinol (P75)
common name (P76) [33] common name (P76)
SMILING Vietnam food code (P77) [34] SMILING Indonesia food code (P78)

Questions

  • Some databases are published in multiple languages. How should we deal with any discrepancies between the translation by the organizations and by Wikidata? (e.g Japanese databases are available in English and Japanese)
    • I think we will be able to present multiple aliases/ multiple values for names. Some of these may conflict, but each will have a reference back to the source. If our group can determine something is incorrect, we can deprecate it.
  • Some databases state that they borrow data from other databases but do not specify exactly which items were borrowed. How should we deal with this? (e.g. Bangladesh FCT 2013)
    • This is my current working model [35]. I am using stated in for the FCT where we found the value and based on for the source they note. How does this seem to you? It is very tricky because sometimes we don't have enough information to decide what to do here.
  • In Wikidata, we can add values without adding references but is it possible to make it a requirement to have at least one reference on WikiFCD?
    • Yes, this is possible.
  • We discussed this before but I forgot the conclusion - do we create different items for the same fruits from different databases (e.g. Apple)?
    • We create a single item for a food item and then statements from all different databases are placed on that item.
  • Is it easy to have Recoin on WikiFCD?
    • Not sure about this. I haven't seen it available for wikibases yet. I'll keep looking.
  • Perhaps we can add some of the identifier properties from WikiProject_Medicine?
    • My current plan is get this data via federated SPARQL queries with Wikidata.