Questions about how Dataverse metadata model aligns with CESSDA Metadata Model

[Julian Gautier]

Hi everyone,

Esra Akdeniz, who's working on GESIS's CESSDA EuroQuestionBank, emailed questions about how the metadata model that ships with Dataverse aligns with the CESSDA Metadata Model. I think others in our community are much more familiar with than I am, particularly about question-level metadata, so we agreed it'll be better to discuss here. I'm pasting Esra's emails below:

"I am writing you in behalf of the CESSDA Metadata Office Project/CESSDA European Question Bank, for which I am taking a closer look at the metadata elements that are covered for Dataverse.

...

For one of our Service Providers, who might consider moving from NESSTAR to Dataverse, I would like to check if Dataverse also covers more detailed metadata on question level, e.g. Question Item Text, Question Item ID, Response Categories/Codes, etc. 

...

I took a closer look at the Dataverse metadata elements compared to our CESSDA Metadata Model (called CMM). CMM is the basis for CESSDA Service Providers and CESSDA tools, a schema all parties agreed on. The good news is that the basic mandatory elements of CMM are covered within Dataverse.

The only exception is that Dataverse does not have a field for study number and does not include multilinguality, which is both crucial for the CESSDA Data Catalogue (CDC) and the CESSDA Euro Question Bank (EQB) (see the mapping attached). For some of our current and future CESSDA Service Providers this would be a significant part of the documentation to include their metadata later into the CESSDA EQB.

I found extensions, such as the Data Curation Tool, that generate additional metadata fields. In this context, the only mention of the "qstn" element from DDI with Dataverse is on the web: https://github.com/IQSS/dataverse/issues/6554. Is this something that CESSDA could use to document questions in Dataverse?"

[Julian Gautier]

I thought I'd try to help clarify two things:

1. The description of the CESSDA EuroQuestionBank at https://www.ukdataservice.ac.uk/about-us/our-rd/cessda-eqb.aspx says that the metadata model is based on DDI-Lifecycle. There are also references to Lifecycle in the mapping/crosswalk that Esra shared (attached in the first post).

Esra, would you be looking into mapping elements from Lifecycle to Codebook, which Dataverse uses? From what I can tell, Codebook doesn't have a Study Number element, but it seems similar to Codebook's Identification Number element, which is what Dataverse maps it's "Other ID" metadata field to.

2. By not including multilinguality, does that mean that Dataverse depositors have no way to indicate that what they've entered in a particular metadata field is a specific language? E.g. if I enter a dataset title in Spanish, I have no way of indicating and Dataverse doesn't try to figure out what language that title is in? I think Amber Leahey brought this up in the GitHub issue at https://github.com/IQSS/dataverse/issues/4632.

[Julian Gautier]

Okay one more thing :)

"I found extensions, such as the Data Curation Tool, that generate additional metadata fields. In this context, the only mention of the "qstn" element from DDI with Dataverse is on the web: https://github.com/IQSS/dataverse/issues/6554. Is this something that CESSDA could use to document questions in Dataverse?"

It's possible to add question-level metadata to existing datasets in a Dataverse repo using the Data Curation Tool. But if we try to migrate dataset metadata into Dataverse using DDI xml files (there's an API endpoint for using DDI xml to create a dataset), and the xml contains question-level metadata, I think Dataverse will ignore it. So I'm not sure if it's possible to migrate dataset metadata that contains question-level metadata into a Dataverse repository.

[Vyacheslav Tikhonov]

Very strange question, it seems some communication problems at GESIS. We have extended Dataverse with CMM metadata schema and testing it for a while, it's a task of DANS at SSHOC project and people from CESSDA community know we're working on it.

Test version of CMM is available here: https://raw.githubusercontent.com/ekoi/speeltuin/master/resources/CMM_Custom_MetadataBlock.tsv

Can you please forward email from Esra Akdeniz to me and Marion?

Best,

Slava 

[Esra Akdeniz]

I received a Dataverse-CMM mapping from Marion, but the mapping does not cover all elements of the CMM. I am not involved in the SSHOC project and only heard from it at the SSHOC Dataverse webinar.

[Vyacheslav Tikhonov]

We're working on the flexible support of metadata schema so if you're missing some fields it's not necessary to map to the existent schema. 

Just add new fields to some Google Spreadsheet and those metadata fields will to be available in Dataverse, there is also support of ANY external controlled vocabularies for specific fields is coming:

https://drive.google.com/file/d/17fQwtTYK91zLpaPbPH3ZMdpAugaA8w7V/view

We'll be more than happy to give a demonstration.

Best,

Slava

[Julian Gautier]

This is great news. Is this work on "the flexible support of metadata schema" being done as part of the metadata working group proposed during the last Dataverse Community Meeting? Or is it something the working group should be aware of if/when it's formed?

Also, the file https://drive.google.com/file/d/17fQwtTYK91zLpaPbPH3ZMdpAugaA8w7V/view isn't publicly viewable. Should it be?

[Vyacheslav Tikhonov]

I think the metadata working group should know about it, indeed. I've made the current version of CMM metadata schema public, please check 

https://drive.google.com/file/d/1hNM7Kz1y-IBAzj_b8F9igm-BMekEatld/view?usp=sharing

Best,

Slava

[Amber Leahey]

Hi Slava, is this to add CESSDA metadata as a custom metadata block? At the dataset level this makes sense. 

It would neat to explore a way to define variable-level metadata as a custom metadata block, especially in the context of migrating metadata but also editing/ writing metadata using the Data Curation Tool. 

BTW you can test the Data Curation Tool in our demo instance (https://demodv.scholarsportal.info/dataverse.xhtml)  by uploading a new dataset with defined variables. More work is being scoped out for this to support editing labels and catagories. 

Best, 

Amber

[Youn Noh]

I downloaded the mapping included as an attachment above and noted that there wasn't a mapping for Principal Investigator. I'm trying to map an element for "Owner" to Dataverse metadata and don't see anything suitable. Is this something that has been considered, or is there an existing element I should use? In DDI-C, I think it might map to citation/rspStmt/AuthEnty and in DDI-L, to citation/creator. Thanks. Youn

[Youn Noh]

Just a note that contributorType in Citation Metadata with the value set to Rights Holder seemed to be an appropriate mapping for Owner ... Thanks.