SEAD needs to collect some information about datasets it stores and preserves to provide quality services and satisfy requirements of institutional repositories. Minimal metadata is the minimum amount of information that SEAD will collect.

There are several ways to collect metadata (a combination will most likely work best:

• require data submitters to provide it
• automatically harvest it
• manually create metadata by working with researchers and external sources

Minimal metadata can be collected when data are first uploaded into SEAD ACR (active data) or then they are marked for publication (final data). When it's active, creator information is probably sufficient. When it's getting finalized, we could ask (or harvest) the following, which will later be mapped onto FGDC or any other standard accepted in a particular domain:

• Main dataset creators (these people will be identified as authors in a data citation)
• Other contributors (these people can be acknowledged or referenced somewhere in additional fields)
• Contact information for questions
• Title of the dataset
• Was data part of a published research? If yes, provide a citation or link.
• Purpose of collecting these data
• Topical keywords that describe this dataset (ideally, we should have both - user-generated and a drop-down list based on domains)
• When were the data collected / created? (can be a date or a period)
• Where the data were collected? (N/A, geographic name, or coordinates)
• What methods were used to gather data? (should be a drop-down list)
• Choice of rights management (we can provide a link to choosing a CC license http://creativecommons.org/choose/ or let people choose from a similar dropdown list that has options of what is allowed for the data, e.g., modify, share, etc.)

For files within a dataset we may need to know software that was used to create the file (important for meaningful files, can be skipped for auxiliary files).