How to Create Effective Metadata
Creating effective metadata can be difficult. How do you know when to use metadata and when not to use metadata? and is meta-metadata ever a good thing?
To Meta or Not to Meta??
Imagine a picture, for arguments sake lets say it is of a beach, at sunset, the beach is rocky and the tide is just starting to turn. *Click*, right now that that is on the digital camera lets take it back to the computer and start tinkering with the data.
Now in a typical situation the camera records some metadata when you take the picture, usually the time and date, the model and serial number of the camera is usually also recorded. If the camera supports it you might even have a longitude and latitude of where the camera was when the picture was taken. This is all very well, but apart from the date and time the rest of the cameras automatic metadata is useless in a data warehouse scenario.
If I search for this image I want to be able to look up “rocky beach” or “sunset” or maybe something like “inspirational photography, I might even want to find “pictures taken by Jamie”. This is all metadata which must be added by humans, in the web 2.0 world they are called “tags”. The same strategy applies to any data.
There are obviously limits to the metadata you should add, e.g. the tag “sun” may be a bit too general and of no real use since the majority of people opt for a long tailed search strategy.
So, if we take my example image what metadata could we come up with?
- Sunset
- Beach-Rocky
- Photographer-Jamie
- Exact Location - Date and Time Note: These would not be considered tags, but they are useful when organising data
You may have noticed that I conventionally tag in the form General-Expanded e.g. Beach-Type. This isn’t the only convention, others prefer to tag full descriptions e.g. RockyBeach or Rocky-Beach. Because metadata is subjective it depends entirely on how you think the information should be represented. I would of course recommend my approach as it allows for structured parsing and means that data can easily be categorised e.g. Beach ยป Rocky
Meta-MetaData?
Meta-Metadata is data about metadata. This may seem like a lot of redundancy but it can actually be quite useful, for example if someone in my data collection tagged a photograph as Beach-Bed this would contradict my metadata design i.e. you cannot have a beach made of beds or at least a natural one. Therefore I may want to create a file that details acceptable expanded definitions of the Beach tag.