Re: Bioschemas.org to define biodiversity-related markup

Hi Mélanie, hi all,

To go a bit further I've tried to somewhat extend the example I've 
initiated. There it is: 
https://github.com/frmichel/taxref-ld/tree/master/bioschemas-org
The README gives details as to how the example file is organized, and 
more importantly it lists some of the issues and questions that we shall 
have to tackle if we officially start the group.

@Alasdair, Carole, Rafael: as discussed in the thread, at some point it 
shall be beneficial to to invite people from EoL and TDWG. Is there some 
sort of "official" channel for the community to do that?

Have a nice week-end,
     Franck.

Le 17/11/2017 à 10:19, Melanie Courtot a écrit :
> Hi Frank, all,
>
> On 16/11/2017 09:37, Franck Michel wrote:
>> Hi Meanie, hi all,
>>
>> EoL provides an API that returns species descriptions as JSON-LD 
>> based on schemas.org. Beluga example: http://eol.org/api/traits/328541
>> It is unclear who consumes this data, but at least, as you already 
>> saw, they embed it at the end of their own web pages such as 
>> http://eol.org/pages/328541/data.
> BioSamples does the same - an API to retrieve JSON and we embed it in 
> our webpages for crawler as well.
>>
>> As you also noticed, the JSON-LD they provide is not valid. I didn't 
>> know about that EOL Github issue, but I recently discussed it with 
>> Rod Page from the Biodiversity Information Standards (aka TDWG), who 
>> replied on the Github issue. The Google structured data testing tool 
>> gives more details on that: https://frama.link/xJm0AAto
>> Besides, other errors are not reported (well, I think these are 
>> errors): property scienfiticName without any namespace is invalid, 
>> that should be dwc:scientificName since this does not exist in 
>> schema.org. Same issue for vernacularName, traits, units...
>>
>> But whatever, this JSON-LD has lots of issues, but it's a start. 
>
> Yes. Only mentioned the tweaks in case someone wanted to give it a try 
> as well.
>
>> The assumption is that there is some sort of specific (one-to-one) 
>> agreement between EoL and Google, and that Google harvests this data 
>> despite the invalid JSON-LD. But I have no confirmation of that
>
> It'd be interesting to clarify this. It seems a little bit counter 
> intuitive that EoL would mark their pages up with JSON for Google to 
> read it but then Google couldn't do so without a special adapter? 
> We're probably missing a piece of the story.
>>
>> > - the measurement type points to 
>> http://purl.obolibrary.org/obo/VT_0001256, which is body length. The 
>> schema.org/predicate value is also "body length (VT)". How is this 
>> understood and displayed as Length on the Google result?
>> - Similar question for the actual value and units, which are 
>> "4249.83" and "mm" respectively. Is Google doing some sort of unit 
>> conversion/roundup for display?
>>
>> Good question. Typically about the unit "mm":
>> - "units": "mm" => there is no such thing as http://schema.org/units
>> - "dwc:measurementUnit": "http://purl.obolibrary.org/obo/UO_0000016" 
>> => this seems to be the only reliable property, but then Google knows 
>> the Darwin Core vocabulary and interprets it.
>> My assumption is that Google performs some treatment on the values. 
>> Possibly, they developed a specific connector to cope with EoL 
>> JSON-LD and translate this body size to "4.2 m".
>> Besides, the snippet mentions "4.2 m *(Adult)*", so they also 
>> presumably consider this property:
>> eol:traitUri"http://eol.org/resources/704/measurements/adultheadbodylen27"
>> to know that this is the size of an adult.
>>
>> With proper Bioschemas.org profiles, I think we could annotate pages 
>> from many other institutions, such as the Beluga page 
>> <https://inpn.mnhn.fr/espece/cd_nom/60932?lg%3Den> on the french 
>> National Museum of Natural History, and in turn, enable search 
>> engines to harvest data from complimentary pages and produce mashups 
>> of related pages, etc.
> That sounds like a great idea and entirely within the scope of Bioschemas.
>>
>> At this point, I think we should involve people from EoL, and from 
>> the TDWG community (Rod Page would certainly be of great added value 
>> in this respect). What do you think? Is there a procedure for 
>> inviting people "officially"?
> I think we could benefit from their experience indeed; it seems they 
> were able to deploy markup, add additional properties and then get 
> this to be interpreted by Google which seems to match our use case 
> pretty well!
> I +1'd the issue at 
> https://github.com/BioSchemas/specifications/issues/115
>
> Cheers,
> Melanie
>
>
>
>
>>
>> Franck.
>>
>>
>> Le 15/11/2017 à 17:57, Melanie Courtot a écrit :
>>> Hi Frank,
>>>
>>> This looks really interesting, thanks for bringing it up. I was 
>>> trying to find out how the interaction between EoL and schema.org 
>>> was working and am wondering if you (or someone else!) could shed 
>>> some light on this?
>>>
>>> As you suggested in the below, I checked the google beluga 
>>> <https://www.google.fr/search?dcr=0&ei=ml74WajPMMzWUabjqvAF&q=beluga&oq=beluga&gs_l=psy-ab.3...19519.20929.0.20945.6.3.0.0.0.0.93.93.1.1.0....0...1.1.64.psy-ab..5.1.92...0j0i131k1.0.AGNziTItYzc> 
>>> search result and do see the line "Length: 4.2 m (Adult) 
>>> Encyclopedia of Life"
>>>
>>> If I try to find where that info comes from, and head to EoL, I can 
>>> reach the page http://eol.org/pages/328541/overview, and follow the 
>>> "see all traits" link to http://eol.org/pages/328541/data which 
>>> contains the JSON-LD.
>>>
>>> I trimmed it down to extract the relevant bit, updated the id to be 
>>> a string as per https://github.com/EOL/tramea/issues/352, and pasted 
>>> it in the JSON playground mostly to make sure it was working as 
>>> expected: http://tinyurl.com/yadam6nj
>>>
>>> I am missing the link of how the following happens:
>>> - the measurement type points to 
>>> http://purl.obolibrary.org/obo/VT_0001256, which is body length. The 
>>> schema.org/predicate value is also "body length (VT)". How is this 
>>> understood and displayed as Length on the Google result?
>>> - Similar question for the actual value and units, which are 
>>> "4249.83" and "mm" respectively. Is Google doing some sort of unit 
>>> conversion/roundup for display?
>>> - Trophic level on EoL is "carnivore", but Google displays "Carnivorous"
>>> etc
>>>
>>> Or am I looking at the wrong source for the markup?
>>>
>>> Cheers,
>>> Melanie
>>>
>>>
>>>
>>>
>>>
>>>
>>> On 10/11/2017 15:17, Franck Michel wrote:
>>>> Dear all,
>>>>
>>>> I've just joined the Bioschemas.org community following some 
>>>> discussions I had with Alasdair Gray whom I met at ISWC in Vienna, 
>>>> and I'd like to start a new discussion thread.
>>>>
>>>> So, just to start, a few words about me. I'm a CNRS research 
>>>> engineer, I work at the I3S laboratory in France, in particular 
>>>> with the Wimmics research team led by Fabien Gandon. I'm currently 
>>>> involved in some activities related to the publication of taxonomic 
>>>> information as Linked Data [1]. In this context, I've met the 
>>>> Biodiversity Information Standards community (TDWG) that is 
>>>> increasingly considering SW standards, LD publication and web pages 
>>>> markup. This is a domain where, I think, it would be relevant for 
>>>> Bioschemas.orgto get involved.
>>>>
>>>> There exist lots of web portals reporting observations, traits and 
>>>> other data about all sorts of living organisms. Encyclopedia of 
>>>> Life <http://eol.org/> (EoL) and the Global Biodiversity 
>>>> Information Facility <https://www.gbif.org/> (GBIF) are some of the 
>>>> most well known. Markup questions are actively considered in this 
>>>> field, for instance EoL web pages embed schemas.org-based JSON-LD 
>>>> descriptions that Google leverages to enrich their snippets: e.g. 
>>>> if you google beluga 
>>>> <https://www.google.fr/search?dcr=0&ei=ml74WajPMMzWUabjqvAF&q=beluga&oq=beluga&gs_l=psy-ab.3...19519.20929.0.20945.6.3.0.0.0.0.93.93.1.1.0....0...1.1.64.psy-ab..5.1.92...0j0i131k1.0.AGNziTItYzc> 
>>>> you shall see 'Encyclopedia of Life' mentions in the snippet 
>>>> providing average weight and size data. For now, this seems to be 
>>>> an "individual" initiative between EoL and Google/schemas.org, but 
>>>> it would make sense if this was part of a broader reflection led by 
>>>> Bioschemas.org.
>>>>
>>>> My opinion is that fostering the use of common markup by these 
>>>> portals could be very effective in helping the biodiversity 
>>>> community to discover information and figure out new data 
>>>> integration scenarios.Within Bioschemas.org, we could define 
>>>> profiles to account for biodiversity-related information.Taxonomic 
>>>> registers are used as the backbone of many web portals, apps and 
>>>> databases related to biodiversity, agronomy and agriculture.For 
>>>> instance, EoL and GBIF both rely on the Catalog of Life 
>>>> <http://www.catalogueoflife.org/> taxonomy. Therefore, we could 
>>>> start with the definition of a profile to describe a taxon and the 
>>>> related scientific and vernacular names thereof. Then, this could 
>>>> be extended with the representation of traits (characteristics of 
>>>> biological organisms), observations, occurrence data, conservation 
>>>> status (e.g. endangered) etc. There already exist vocabularies for 
>>>> such data such as the well-adopted Darwin Core terms.
>>>>
>>>> As a quick example, consider the web page describing the common 
>>>> dolphin on the web site of the french Museum of Natural History: 
>>>> https://inpn.mnhn.fr/espece/cd_nom/60878?lg=en. This page could 
>>>> come with a JSON-LD desciption looking like this: 
>>>> https://github.com/frmichel/taxref-ld/blob/master/bioschemas-org-example.json
>>>> This example is naive and very succinct, and there are lots of 
>>>> things to discuss and decide. Besides, I've just registered on the 
>>>> mailing yesterday, so it may not fit with good practices that you 
>>>> guys have already agreed upon. Sorry if this is the case. 
>>>> Nevertheless, my point is basically to bootstrap the discussion and 
>>>> see if the community is willing to endorse this initiative. If this 
>>>> is the case, we should probably involve people from the 
>>>> biodiversity community: Darwin Core experts, EoL/GBIF 
>>>> representatives etc. But that will come in time.
>>>>
>>>> I look forward to further discussions.
>>>> Regards,
>>>>    Franck.
>>>>
>>>> [1] Michel F., Gargominy O., Tercerie S. & Faron-Zucker C. (2017). 
>>>> A Model to Represent Nomenclatural and Taxonomic Information as 
>>>> Linked Data. Application to the French Taxonomic Register, TAXREF. 
>>>> In Proceedings of the 2nd International Workshop on Semantics for 
>>>> Biodiversity (S4BioDiv) co-located with ISWC 2017 vol. 1933. 
>>>> Vienna, Austria. CEUR.
>>>>
>>>> -- 
>>>> signature
>>>>  
>>>> Franck MICHEL
>>>> CNRS research engineer
>>>>  +33 (0)492 96 5004
>>>> franck.michel@cnrs.fr <mailto:franck.michel@cnrs.fr>
>>>>
>>>>  
>>>>
>>>> Université Côte d’Azur, CNRS, *Inria* - I3S - UMR 7271
>>>> 930 route des Colles - Bât. Les Templiers
>>>> BP 145 - 06903 Sophia Antipolis CEDEX - France
>>>> Tel. +33 (0)4 9294 2680, Fax : +33 (0)4 9294 2898
>>>>
>>>
>>
>