Because of the Dataverse object hierarchy, researchers are more likely to split materials for one project/manuscript into multiple DOI-backed deposits that are under a project-specific dataverse (examples for the Texas Data Repository here and here). It could be argued that this inflates dataset counts similar to how Figshare sometimes splits each supplemental file for one manuscript into individual DOIs. Currently, there isn't any cross-walking of an encapsulating dataverse into DataCite (would be interesting to see if 'container' could be used for this). Possible implementations:

• Multi-field deduplication using fields like publication date, author(s), and title. This would probably work for most instances but risks missing outliers such as if related objects are published on different days for some reason or have different author lists.
• Extraction of dataverse information from Dataverse API. This would be more reliable but could still require deduplicating on certain fields in case one dataverse houses materials for several distinct projects and would require querying the Dataverse API for information (more time-intensive, complicated for users who have never had to use the Dataverse API and/or have no need for use otherwise).