Social file sharing database schema

Seriously look at third normal form. I would use surrogate keys with the natural keys implemented as unique keys. You will likely find that author belongs in its own authors table. You may find you have a few tables which are quite similar such as user_content_faves, user_author_faves, user_author_shares. This is normal.

Having a single content table with a content_type column may be appropriate. The content column would need to be capable of storing all the content types.

EDIT: For relationship tables I usually name the join table by concatenating the names of the joined tables, abbreviating as necessary. If there multiple relationships between the two tables, I use one of two options:

• Append the purpose of the relationship (as I did above); or
• Add a type/reason code to the relationship (in which case the type is not needed in the relationship name).

I made the assumption that you would want to track who favorited or shared things. It appears that both the producer (author or user) as well as the content items. Therefore you have users favoriting a producer (user_author_faves now user_user_faves), or a product (user_content_faves). Depending on how you do sharing,

• it could be an attribute on the content, or
• a relationship like user_content_shares, where content is share with a particular user. Re-sharing could be problematic, if you track who shared things, and multiple users share the same content to the same user. Un-sharing re-shared items is problematic if you don't track who did the sharing.

You may want to consider (and set policies for):

• tracking/audit information like when something was done (added, favorited, shared, etc).
• whether to do physical or logical deletes.
• if you do logical deletes how to handle refavoriing or sharing something after a logical delete.

When indexing relationship tables I generally have the primary key consisting of the primary keys of the two tables being joined. A second index with the primary keys reversed, or just the primary key which is the second column in the primary key is usually required. If the relationship between two rows can occur more than once, the column(s) used to differentiate the reason/type and/or timing (since date) of the relationship needs to added to the primary key.

What you're thinking about is either "row modelling" or "entity-attribute-value" (EAV) - See Wikipedia - either of which is a common subject for religious wars. I won't get too far into the pros vs. cons. If you read up on EAV and row modeling you'll see lots of discussion on that.

What I will say is that you should model all of the common predicates using properly normalized tables and rely on EAV (or row modelling) only for the customizable columns (if at all). This is how many highly customizable packaged solutions deal with allowing users to add their own attributes to existing objects. It isn't clean, easy or efficient, but it is probably the best compromise for allowing this particular type of customization when you have no way of reasonably predicting what the customized columns might be at design time.