Response to Issue #4DODS Basic IssuesResponse to Issue #2NOPP Regional Workshop Synthesis Report
Response to Issue #3

Response to Issue #3

What types of semantic metadata will be required and what semantic metadata standard(s) will be used? The focus should be on search and use metadata.

"Search" metadata is the information needed to select a dataset to use (e.g. location and time of measurements), while "use" metadata is information needed to use a dataset (e.g. missing value flags, or units). The two sets overlap, but they are distinct notions.

All of the regions agreed that semantic metadata was vital to the DODS effort. Most of the workshop focus was on use metadata, with relatively little attention paid to search metadata. Workshop attendees emphasized the importance of being able to find the data you need, but the metadata requirements proposed had more to do with making the data you've found easy to use.

There was not a great deal of discussion about how important it is that use metadata be machine-readable--that is, supplied in a consistent format a computer can parse.

Several lengthy discussions dealt with the varieties of use metadata that make datasets valuable, though there was little talk about interoperability with respect to this variety of metadata. One possible implication of this is that machine-level interoperability was not seen as crucial by workshop participants. If someone is going to use data from two different datasets, perhaps they will inevitably do so at the user level rather than at the machine level, if only to maintain a degree of control over the process.

The importance of associating existing data set descriptions--in the FGDC clearinghouse or the GCMD--to data sets was dealt with in at least one of the workshops. This led to a discussion of third party metadata description in general and the value of rendering these accessible via the data access protocol, even when the descriptions and the data do not reside on the same server.

Topping the workshop lists of essential metadata types were:

Names and descriptions
What is meant by the abbreviation of the variable name?
Units and scaling factors
What units are the data delivered in?
Missing values
How are missing values flagged in the dataset?

Several participants commented that a useful metadata collection would have a place for data quality indicators. Suggestions for these took many different forms, including simple flags to indicate accuracy and precision (Gulf Coast workshop), to descriptions of quality control procedures (West Coast), to quality "flags" (Southeast). One suggestion (from the Southeast workshop) was that if a web-based description exists for a dataset, the URL for this description should be in the dataset's data attribute structure (DAS).

Finally, there were several comments about metadata people wished they could have for datasets they would like to use. These suggestions included:

The Southeast and Gulf Coast workshops were explicit in urging that the DODS project accommodate the Z39.50 metadata standard, and do so by working with the FGDC Clearinghouse. They suggested that DODS should be able to use metadata supplied by a third party, such as FGDC, and should be able to do so automatically. The Gulf Coast workshop further recommended that data originators should make use of commercial software packages to assist them in preparing metadata in FGDC format.

One suggestion (from the Southeast workshop) was that support should be provided to those interested in proper descriptions of their datasets.This support could take the form of:

  1. A white paper that would be made available via either the national or a regional DODS website, and
  2. Training sessions offered either at a central site or at individual institutions.

Response to Issue #4DODS Basic IssuesResponse to Issue #2NOPP Regional Workshop Synthesis Report
Response to Issue #3