You may not like this answer, but there is no one size fits all for metadata. The metadata that would best describe your data depends on the nature of your data. However, due to the fact that specialised systems for description (know as schemas) take time and expertise to implement, there are some generalised schemas that have been developed. These schemas include what are considered essential elements to describe any data. There is a trade off between using a generalised metadata schema (limited time and expertise are required, but the description of the data may be inadequate in many respects) and using a specialised schema (requires a greater amount of time and expertise, but results in a much better description of the data).
Dublin Core is the most popular generalised schema. In 1998, the following 15 elements were included in the Dublin Core schema:
Dublin Core has been significantly modified since 1998 as notions of best practice in the Semantic Web have evolved to include the assignment of formal domains and ranges in addition to definitions in natural language. A vast amount of information on past and future planned developments of the schema are available at the Dublin Core Metadata Initiative webpages.
There are many schemas that have been developed for certain types of data, and if your data matches a developed schema, the use of that schema will result in the best metadata for your data. However, many of these schemas are very complex and require a level of expertise to implement that preclude their use by many. Metadata schemas that are in use at the ANU are listed in the box to the right. Other metadata schemas include:
Geospatial metadata (ISO 19115) defines how to describe geographical information and associated services, including contents, spatial-temporal purchases, data quality, access and rights to use. The standard defines more than 400 meta data elements, 20 core elements. An examples of an implementation of ISO19115 is Global Change Master Directory
Metadata is often described as "data about data".
You can think of the metadata, in relation to the data it describes, as being analogous to the abstract or keywords of a paper – it is there to help people find your data and quickly decide if it is what they need. If you want people to find and reuse your data (and therefore help you by citing your work), then it is worth your while making good metadata in order to ‘sell’ your data.
RIF-CS (Registry Interchange Format for Collections and Services) is used by RDA (Research Data Australia). Metadata from the ANU Data Commons is converted into RIF-CS and added to RDA.