Metadata Customization

The Dataverse Software has a flexible data-driven metadata system powered by “metadata blocks” that are listed in the Appendix section of the User Guide. In this section we explain the customization options.


Before you embark on customizing metadata in your Dataverse installation you should make sure you are aware of the modest amount of customization that is available with your Dataverse installation’s web interface. It’s possible to hide fields and make fields required or conditionally required by clicking “Edit” at the Dataverse collection level, clicking “General Information” and making adjustments under “Metadata Fields” as described in the Create a New Dataverse Collection section of the Dataverse Collection Management page in the User Guide.

Much more customization of metadata is possible, but this is an advanced topic so feedback on what is written below is very welcome. The possibilities for customization include:

  • Editing and adding metadata fields

  • Editing and adding instructional text (field label tooltips and text box watermarks)

  • Editing and adding controlled vocabularies

  • Changing which fields depositors must use in order to save datasets (see also Dataset Templates section of the User Guide.)

  • Changing how saved metadata values are displayed in the UI

Generally speaking it is safer to create your own custom metadata block rather than editing metadata blocks that ship with the Dataverse Software, because changes to these blocks may be made in future releases. If you’d like to make improvements to any of the metadata blocks shipped with the Dataverse Software, please open an issue at so it can be discussed before a pull request is made. Please note that the metadata blocks shipped with the Dataverse Software are based on standards (e.g. DDI for social science) and you can learn more about these standards in the Appendix section of the User Guide. If you have developed your own custom metadata block that you think may be of interest to the Dataverse community, please create an issue and consider making a pull request as described in the Version Control section of the Developer Guide.

In current versions of the Dataverse Software, custom metadata are no longer defined as individual fields, as they were in Dataverse Network (DVN) 3.x, but in metadata blocks. The Dataverse Software now ships with a citation metadata block, which includes mandatory fields needed for assigning persistent IDs to datasets, and domain specific metadata blocks. For a complete list, see the Appendix section of the User Guide.

Definitions of these blocks are loaded into a Dataverse installation in tab-separated value (TSV). 1, 2 While it is technically possible to define more than one metadata block in a TSV file, it is good organizational practice to define only one in each file.

The metadata block TSVs shipped with the Dataverse Software are in /tree/develop/scripts/api/data/metadatablocks and the corresponding ResourceBundle property files /tree/develop/src/main/java of the Dataverse Software GitHub repo. Human-readable copies are available in this Google Sheets document but they tend to get out of sync with the TSV files, which should be considered authoritative. The Dataverse Software installation process operates on the TSVs, not the Google spreadsheet.

About the metadata block TSV

Here we list and describe the purposes of each section and property of the metadata block TSV.

  1. metadataBlock

    • Purpose: Represents the metadata block being defined.

    • Cardinality:

      • 0 or more per Dataverse installation

      • 1 per Metadata Block definition

  2. datasetField

    • Purpose: Each entry represents a metadata field to be defined within a metadata block.

    • Cardinality: 1 or more per metadataBlock

  3. controlledVocabulary

    • Purpose: Each entry enumerates an allowed value for a given datasetField.

    • Cardinality: zero or more per datasetField

Each of the three main sections own sets of properties:

#metadataBlock properties



Allowed values and restrictions


A user-definable string used to identify a #metadataBlock

• No spaces or punctuation, except underscore.

• By convention, should start with a letter, and use lower camel case 3

• Must not collide with a field of the same name in the same or any other #datasetField definition, including metadata blocks defined elsewhere. 4


If specified, this metadata block will be available only to the Dataverse collection designated here by its alias and to children of that Dataverse collection.

Free text. For an example, see custom_hbgdki.tsv.


Acts as a brief label for display related to this #metadataBlock.

Should be relatively brief. The limit is 256 character, but very long names might cause display problems.


Associates the properties in a block with an external URI. Properties will be assigned the global identifier blockURI<name> in the OAI_ORE metadata and archival Bags

The citation #metadataBlock has the blockURI which assigns a default global URI to terms such as

#datasetField (field) properties



Allowed values and restrictions


A user-definable string used to identify a #datasetField. Maps directly to field name used by Solr.

• (from The internal DDI-like name, no spaces, etc.

• (from Solr) Field names should consist of alphanumeric or underscore characters only and not start with a digit. This is not currently strictly enforced, but other field names will not have first class support from all components and back compatibility is not guaranteed. Names with both leading and trailing underscores (e.g. _version_) are reserved.

• Must not collide with a field of the same same name in another #metadataBlock definition or any name already included as a field in the Solr index.


Acts as a brief label for display related to this #datasetField.

Should be relatively brief.


Used to provide a description of the field.

Free text


A string to initially display in a field as a prompt for what the user should enter.

Free text


Defines the type of content that the field, if not empty, is meant to contain.

• none • date • email • text • textbox • url • int • float • See below for fieldtype definitions


Controls the sequence in which the fields are displayed, both for input and presentation.

Non-negative integer.


Controls how the content is displayed for presentation (not entry). The value of this field may contain one or more special variables (enumerated below). HTML tags, likely in conjunction with one or more of these values, may be used to control the display of content in the web UI.

See below for displayFormat variables


Specify whether this field is available in advanced search.

TRUE (available) or FALSE (not available)


Specify whether the possible values of this field are determined by values in the #controlledVocabulary section.

TRUE (controlled) or FALSE (not controlled)


Specify whether this field is repeatable.

TRUE (repeatable) or FALSE (not repeatable)


Specify whether the field is facetable (i.e., if the expected values for this field are themselves useful search terms for this field). If a field is “facetable” (able to be faceted on), it appears under “Browse/Search Facets” when you edit “General Information” for a Dataverse collection. Setting this value to TRUE generally makes sense for enumerated or controlled vocabulary fields, fields representing identifiers (IDs, names, email addresses), and other fields that are likely to share values across entries. It is less likely to make sense for fields containing descriptions, floating point numbers, and other values that are likely to be unique.

TRUE (controlled) or FALSE (not controlled)

displayoncreate 5

Designate fields that should display during the creation of a new dataset, even before the dataset is saved. Fields not so designated will not be displayed until the dataset has been saved.

TRUE (display during creation) or FALSE (don’t display during creation)


For primitive fields, specify whether or not the field is required.

For compound fields, also specify if one or more subfields are required or conditionally required. At least one instance of a required field must be present. More than one instance of a field may be allowed, depending on the value of allowmultiples.

For primitive fields, TRUE (required) or FALSE (optional).

For compound fields:

• To make one or more subfields optional, the parent field and subfield(s) must be FALSE (optional).

• To make one or more subfields required, the parent field and the required subfield(s) must be TRUE (required).

• To make one or more subfields conditionally required, make the parent field FALSE (optional) and make TRUE (required) any subfield or subfields that are required if any other subfields are filled.


For subfields, specify the name of the parent or containing field.

• Must not result in a cyclical reference.

• Must reference an existing field in the same #metadataBlock.


Specify the name of the #metadataBlock that contains this field.

• Must reference an existing #metadataBlock.

• As a best practice, the value should reference the #metadataBlock in the current definition (it is technically possible to reference another existing metadata block.)


Specify a global URI identifying this term in an external community vocabulary.

This value overrides the default (created by appending the property name to the blockURI defined for the #metadataBlock)

For example, the existing citation #metadataBlock defines the property named ‘title’ as - i.e. indicating that it can be interpreted as the Dublin Core term ‘title’

#controlledVocabulary (enumerated) properties



Allowed values and restrictions


Specifies the #datasetField to which #datasetField to which this entry applies.

Must reference an existing #datasetField. As a best practice, the value should reference a #datasetField in the current metadata block definition. (It is technically possible to reference an existing #datasetField from another metadata block.)


A short display string, representing an enumerated value for this field. If the identifier property is empty, this value is used as the identifier.

Free text


A string used to encode the selected enumerated value of a field. If this property is empty, the value of the “Value” field is used as the identifier.

Free text


Control the order in which the enumerated values are displayed for selection.

Non-negative integer.

FieldType definitions




Used for compound fields, in which case the parent field would have no value and display no data entry control.


A date, expressed in one of three resolutions of the form YYYY-MM-DD, YYYY-MM, or YYYY.


A valid email address. Not indexed for privacy reasons.


Any text other than newlines may be entered into this field.


Any text may be entered. For input, the Dataverse Software presents a multi-line area that accepts newlines. While any HTML is permitted, only a subset of HTML tags will be rendered in the UI. See the Supported HTML Fields section of the Dataset + File Management page in the User Guide.


If not empty, field must contain a valid URL.


An integer value destined for a numeric field.


A floating point number destined for a numeric field.

displayFormat variables

These are common ways to use the displayFormat to control how values are displayed in the UI. This list is not exhaustive.




The displayFormat is left blank for primitive fields (e.g. subtitle) and fields that do not take values (e.g. author), since displayFormats do not work for these fields.


The value of the field (instance level).


The name of the field (class level).


For displaying emails.

<a href=”#VALUE”>#VALUE</a>

For displaying the value as a link (if the value entered is a link).

<a href=’URL/#VALUE’>#VALUE</a>

For displaying the value as a link, with the value included in the URL (e.g. if URL is, and the value entered is 1001, the field is displayed as 1001 (hyperlinked to

<img src=”#VALUE” alt=”#NAME” class=”metadata-logo”/><br/>

For displaying the image of an entered image URL (used to display images in the producer and distributor logos metadata fields).




Appends and/or prepends characters to the value of the field. e.g. if the displayFormat for the distributorAffiliation is (#VALUE) (wrapped with parens) and the value entered is University of North Carolina, the field is displayed in the UI as (University of North Carolina).




Displays the character (e.g. semicolon, comma) between the values of fields within compound fields. For example, if the displayFormat for the compound field “series” is a colon, and if the value entered for seriesName is IMPs and for seriesInformation is A collection of NMR data, the compound field is displayed in the UI as IMPs: A collection of NMR data.

Metadata Block Setup

Now that you understand the TSV format used for metadata blocks, the next step is to attempt to make improvements to existing metadata blocks or create entirely new metadata blocks. For either task, you should have a Dataverse Software development environment set up for testing where you can drop the database frequently while you make edits to TSV files. Once you have tested your TSV files, you should consider making a pull request to contribute your improvement back to the community.

Exploring Metadata Blocks

In addition to studying the TSV files themselves you might find the following highly experimental and subject-to-change API endpoints useful to understand the metadata blocks that have already been loaded into your Dataverse installation:

You can get a dump of metadata fields (yes, the output is odd, please open a issue) like this:

curl http://localhost:8080/api/admin/datasetfield

To see details about an individual field such as “title” in the example below:

curl http://localhost:8080/api/admin/datasetfield/title

Setting Up a Dev Environment for Testing

You have several options for setting up a dev environment for testing metadata block changes:

To get a clean environment in Vagrant, you’ll be running vagrant destroy. In Docker, you’ll use docker rm. For a full dev environment or AWS installation, you might find rebuild and related scripts at scripts/deploy/ useful.

Editing TSV files

Early in Dataverse Software 4.0 development, metadata blocks were edited in the Google spreadsheet mentioned above and then exported in TSV format. This worked fine when there was only one person editing the Google spreadsheet but now that contributions are coming in from all over, the TSV files are edited directly. We are somewhat painfully aware that another format such as XML might make more sense these days. Please see for a discussion of non-TSV formats.

Please note that metadata fields share a common namespace so they must be unique. The following curl command will print the list of metadata fields already available in the system:

curl http://localhost:8080/api/admin/index/solr/schema

We’ll use this command again below to update the Solr schema to accomodate metadata fields we’ve added.

Loading TSV files into a Dataverse Installation

A number of TSV files are loaded into a newly-installed Dataverse installation, becoming the metadata blocks you see in the UI. For the list of metadata blocks that are included with the Dataverse Software out of the box, see the Appendix section of the User Guide.

Along with TSV file, there are corresponding ResourceBundle property files with key=value pair here. To add other language files, see the Configuration for JVM Options section, and add a file, for example: “” to the path you specified for the JVM option, and then restart the app server.

If you are improving an existing metadata block, the Dataverse Software installation process will load the TSV for you, assuming you edited the TSV file in place. The TSV file for the Citation metadata block, for example, can be found at scripts/api/data/metadatablocks/citation.tsv. If any of the below mentioned property values are changed, corresponding ResourceBundle property file has to be edited and stored under location

  • name, displayName property under #metadataBlock

  • name, title, description, watermark properties under #datasetfield

  • DatasetField, Value property under #controlledVocabulary

If you are creating a new custom metadata block (hopefully with the idea of contributing it back to the community if you feel like it would provide value to others), the Dataverse Software installation process won’t know about your new TSV file so you must load it manually. The script that loads the TSV files into the system is scripts/api/ and contains a series of curl commands. Here’s an example of the necessary curl command with the new custom metadata block in the “/tmp” directory.

curl http://localhost:8080/api/admin/datasetfield/load -H "Content-type: text/tab-separated-values" -X POST --upload-file /tmp/new-metadata-block.tsv

To create a new ResourceBundle, here are the steps to generate key=value pair for the three main sections:

#metadataBlock properties value of name property from #metadatablock)

metadatablock.displayName=(the value of displayName property from #metadatablock)


metadatablock.displayName=Citation Metadata

#datasetField (field) properties

datasetfieldtype.(the value of name property from #datasetField).title=(the value of title property from #datasetField)

datasetfieldtype.(the value of name property from #datasetField).description=(the value of description property from #datasetField)

datasetfieldtype.(the value of name property from #datasetField).watermark=(the value of watermark property from #datasetField)



datasetfieldtype.title.description=Full title by which the Dataset is known.

datasetfieldtype.title.watermark=Enter title…

#controlledVocabulary (enumerated) properties

controlledvocabulary.(the value of DatasetField property from #controlledVocabulary).(the value of Value property from #controlledVocabulary)=(the value of Value property from #controlledVocabulary)

Since the Value property from #controlledVocabulary is free text, while creating the key, it has to be converted to lowercase, replace space with underscore, and strip accents.


controlledvocabulary.subject.agricultural_sciences=Agricultural Sciences

controlledvocabulary.language.marathi_(marathi)=Marathi (Maru0101u1E6Dhu012B)

Enabling a Metadata Block

Running a curl command like “load” example above should make the new custom metadata block available within the system but in order to start using the fields you must either enable it from the UI (see General Information section of Dataverse Collection Management in the User Guide) or by running a curl command like the one below using a superuser API token. In the example below we are enabling the “journal” and “geospatial” metadata blocks for the root Dataverse collection:

curl -H "X-Dataverse-key:$API_TOKEN" -X POST -H "Content-type:application/json" -d "[\"journal\",\"geospatial\"]" http://localhost:8080/api/dataverses/:root/metadatablocks

Updating the Solr Schema

Once you have enabled a new metadata block you should be able to see the new fields in the GUI but before you can save the dataset, you must add additional fields to your Solr schema.

An API endpoint of your Dataverse installation provides you with a generated set of all fields that need to be added to the Solr schema configuration, including any enabled metadata schemas:

curl "http://localhost:8080/api/admin/index/solr/schema"

You can use to easily add these to the Solr schema you installed for your Dataverse installation.

The script needs a target XML file containing your Solr schema. (See the Prerequisites section of the Installation Guide for a suggested location on disk for the Solr schema file.)

You can either pipe the downloaded schema to the script or provide the file as an argument. (We recommended you to take a look at usage output of -h)

Example usage of
curl "http://localhost:8080/api/admin/index/solr/schema" | /usr/local/solr/server/solr/collection1/conf/schema.xml

You will need to reload your Solr schema via an HTTP-API call, targeting your Solr instance:

curl "http://localhost:8983/solr/admin/cores?action=RELOAD&core=collection1"

You can easily roll your own little script to automate the process (which might involve fetching the schema bits from some place else than your Dataverse installation).

Please note that reconfigurations of your Solr index might require a re-index. Usually release notes indicate a necessary re-index, but for your custom metadata you will need to keep track on your own.

Please note also that if you are going to make a pull request updating conf/solr/8.11.1/schema.xml with fields you have added, you should first load all the custom metadata blocks in scripts/api/data/metadatablocks (including ones you don’t care about) to create a complete list of fields. (This might change in the future.)

Reloading a Metadata Block

As mentioned above, changes to metadata blocks that ship with the Dataverse Software will be made over time to improve them and release notes will sometimes instruct you to reload an existing metadata block. The syntax for reloading is the same as loading. Here’s an example with the “citation” metadata block:

curl http://localhost:8080/api/admin/datasetfield/load -H "Content-type: text/tab-separated-values" -X POST --upload-file citation.tsv

Great care must be taken when reloading a metadata block. Matching is done on field names (or identifiers and then names in the case of controlled vocabulary values) so it’s easy to accidentally create duplicate fields.

The ability to reload metadata blocks means that SQL update scripts don’t need to be written for these changes. See also the SQL Upgrade Scripts section of the Developer Guide.

Using External Vocabulary Services

The Dataverse software has a mechanism to associate specific fields defined in metadata blocks with a vocabulary(ies) managed by external services. The mechanism relies on trusted third-party Javascripts. The mapping from field type to external vocabulary(ies) is managed via the :CVocConf setting.

This functionality is considered ‘experimental’. It may require significant effort to configure and is likely to evolve in subsequent Dataverse software releases.

The effect of configuring this mechanism is similar to that of defining a field in a metadata block with ‘allowControlledVocabulary=true’:

  • Users are able to select from a controlled list of values.

  • Values can be shown in any language the term has been defined in.

In general, the external vocabulary support mechanism may be a better choice for large vocabularies, hierarchical/structured vocabularies, and/or vocabularies managed by third-parties. In addition, the external vocabulary mechanism differs from the internal controlled vocabulary mechanism in several ways that may make it a preferred option:

  • the machine-readable URI form of a vocabulary is stored in the Dataverse database and can be included in exported metadata files.

  • vocabulary mappings can be changed without changing the metadata block, making it possible for different Dataverse installations to use different vocabularies in the same field.

  • mappings can associate a field with more than one vocabulary.

  • mappings can be configured to also allow custom/free-text entries as well as vocabulary values.

  • mappings can be configured for compound fields and a user’s selection of a given vocabulary value can be used to fill in related child fields (e.g. selection of a keyword could fill in a vocabulary name field as well).

  • removing a mapping does not affect stored values (the field would revert to allowing free text).

The specifics of the user interface for entering/selecting a vocabulary term and how that term is then displayed are managed by third-party Javascripts. The initial Javascripts that have been created provide auto-completion, displaying a list of choices that match what the user has typed so far, but other interfaces, such as displaying a tree of options for a hierarchical vocabulary, are possible. Similarly, existing scripts do relatively simple things for displaying a term - showing the term’s name in the appropriate language and providing a link to an external URL with more information, but more sophisticated displays are possible.

Scripts supporting use of vocabularies from services supporting the SKOMOS protocol (see and retrieving ORCIDs (from are available (Custom scripts can also be used and community members are encouraged to share new scripts through the dataverse-external-vocab-support repository.)

Configuration involves specifying which fields are to be mapped, whether free-text entries are allowed, which vocabulary(ies) should be used, what languages those vocabulary(ies) are available in, and several service protocol and service instance specific parameters. These are all defined in the :CVocConf setting as a JSON array. Details about the required elements as well as example JSON arrays are available at, along with an example metadata block that can be used for testing. The scripts required can be hosted locally or retrieved dynamically from (similar to how dataverse-previewers work).

Please note that in addition to the :CVocConf described above, an alternative is the :ControlledVocabularyCustomJavaScript setting.

Tips from the Dataverse Community

When creating new metadata blocks, please review the Text section of the Style Guide, which includes guidance about naming metadata fields and writing text for metadata tooltips and watermarks.

If there are tips that you feel are omitted from this document, please open an issue at and consider making a pull request to make improvements. You can find this document at

Alternatively, you are welcome to request “edit” access to this “Tips for Dataverse Software metadata blocks from the community” Google doc:

The thinking is that the tips can become issues and the issues can eventually be worked on as features to improve the Dataverse Software metadata system.




Although the structure of the data, as you’ll see below, violates the “Each record must have the same number of fields” tenet of TSV



These field names are added to the Solr schema.xml and cannot be duplicated. See “Editing TSV files” for how to check for duplication.


“displayoncreate” was “showabovefold” in Dataverse Software <=4.3.1 (see #3073) but parsing is done based on column order rather than name so this only matters to the person reading the TSV file.