Earlier in this series, I introduced four common challenges that I frequently encounter in geospatial integration projects. After exploring the importance of knowing your data in the previous post, I now turn to the second challenge: creating standards-based metadata.
In one of my previous projects, when I asked colleagues if they had metadata, many answered “yes”. However, I soon discovered that these “metadata” were files in ad-hoc formats, containing only discovery metadata, or software-generated outputs with statistics about the file or underlying model.
In the geospatial domain, people are generally more familiar with the concept of standards-based metadata—metadata that follows established standards such as ISO 19115, INSPIRE, Dublin Core, or DCAT—than they are in many other scientific disciplines.
Metadata is a cornerstone of the F.A.I.R. principles, helping make datasets findable and accessible. However, if metadata does not follow a standard, it is neither interoperable nor reusable.
Why Standards-Based Metadata Matters
Creating standards-based metadata, however, still seems to be a daunting task for many people. This may be due to a lack of tools that support metadata authoring, combined with the perception that metadata itself is complicated.
It also does not help that many people don’t perceive metadata as particularly useful, leaving them with little motivation to invest time in creating it.
I think choosing the right metadata schema is incredibly important here. There is a “sweet spot” between creating something that is still useful for the project while not overloading users with too many fields to fill.
In some cases, discovery metadata alone could be sufficient—for example, a few fields like in OGC API – Records atomic unit. In other cases, it should be possible to extract some information automatically, leaving users to complete only a handful of fields.
The goal should be to ask for as little manual input as possible while ensuring that the information provided is accurate and meaningful.
After all, the only thing worse than having no metadata is having poor-quality metadata. When a system ingests bad quality metadata, it fails to provide useful information, reinforcing the perception that metadata itself has little value.
Making Metadata Easier to Create
Unfortunately, the ecosystem of tools for authoring and viewing metadata is still not as “rich” as the data ecosystem—a gap that I believe we should address.
However, we can still help users create standards-based metadata easily, by avoiding workflows that require them to submit XML or JSON files directly.
One practical approach is to provide a survey form and ask users a small number of questions about the data, making clear that they should fill out a survey for each new dataset.
Where users manage many datasets with repetitive information, a spreadsheet can be more efficient, allowing information to be copied and pasted across multiple records. An ETL can then validate these responses, transform them into machine-readable metadata, and eventually ingest them into the system.
The eMOTIONAL Cities project offered users several ways to submit metadata into the Spatial Data Infrastructure (SDI), ranging from human-friendly interfaces, such as survey forms, to machine-friendly approaches (e.g., POST requests).
Changing the Perception of Metadata
Although we can, and should, increase the number of metadata clients and create workflows that help users create metadata with existing tools, the perception of metadata as “not relevant” will remain an obstacle to the proliferation of metadata.
As we see with data itself, people are willing to spend a significant amount of time and effort in producing high-quality outputs. We need to foster the same mindset for metadata, and, in my experience, that can only happen through education and outreach, by demonstrating its real value.
