Semantic Overflow

Question

2

Can you advise me concerning the wisdom or otherwise of storing entire documents as the objects of triples?

I'm building a CMS-like application, and need to store thousands of XHTML-formatted documents somewhere on the server. I'm using Allegrograph to create and store semantic meta-data about the documents; but what about the documents themselves? Are there any disadvantages to storing them as xsd:string values in the triple-store? Is it considered better practice to store the documents elsewhere - flat files in a directory? - and reference them from triples?

flag	asked Jul 13 at 5:42 dpinn 119●6

cygri · Accepted Answer · 2010-07-13 07:58:09Z UTC

Advantages to storing your files as files in the file system:

Easier development and debugging
Easier data management, e.g., for backups
Probably faster (but may not be a problem at your scale) because it keeps the “big blobs” out of the triple store
Avoids problems with round tripping of Unicode characters etc

Advantages to storing your files as literals in the triple store:

Uniform access to data and metadata
Easier to keep data and metadata in sync
Versioning is more easily implemented

Personally I'd keep them as files.

ldodds · Answer 2 · 2010-07-13 09:27:33Z UTC

While there may be advantages in storing your documents in your triple store, e.g. only one data store to manage, interface to code against, etc. I'd argue that its simpler and more flexible to store your files in the file system, perhaps making them accessible over HTTP with a link to the document stored as a triple in your store.

As your application develops you'll likely to want to add features that relate specifically to the documents (e.g. editing, full-text indexing, validation, etc) and this will be easier to do if you're not constrained by having to integrate all that with a triple store.

Should I store documents as string values in my triple store?

2 Answers