What's the most common method for exploding a string of tags?

Question

I'm not really concerned about the implementation so much as what the most common method would be. Our users aren't very techy so we get a lot of tags that don't exactly follow directions. I'd rather conform to a greater standard than try to make it up myself.

Should we even allow different methods or just accept one delimeter like ", "?

An extreme example string might be:

dogs, bad_mojo + help-me yawning,, "who's there?", red shoes

Delicious turns that string into these tags:

• dogs
• bad_mojo
• +
• help-me
• yawning,
• "who's
• there?"
• red
• shoes

That seems fairly smart but you can see how the useless junk can get in there.

Answer 1

It really is up to you. There are plenty of good solutions. You can use spaces or a symbol as the delimiter. I recommend picking on and making sure it is obvious to the user how they should do it.

If you use spacing then you will need to make sure that users know to join words using hyphens or something similar. Or you can allow multi-word tags by allowing them to use quotes.

I wouldn't go all out allowing any type of separator, but as long as you decide if tags can be more than one word you'll be able to make the right choice for your circumstances.

Answer 2

I would split dogs, bad_mojo + help-me yawning,, "who's there?", red shoes into the tags:

• dogs
• bad mojo
• help me
• yawning
• who's there?
• red shoes

I would accept + in C++, but I wouldn't accept any single character tags.

If you want to use SO-style tags, you could process these to:

• dogs
• bad-mojo
• help-me
• yawning
• whos-there
• red-shoes

Mechanical turk seems to be a good route to sort-out all the tags.