[roywiggins]

You could do something like the Rosetta disk, where you just physically engrave the glyphs.

http://rosettaproject.org/disk/concept/

Of course then you need to work out what the language is actually saying, but plain "glyph retrieval" can be done with a desktop microscope and some time. An alien that operates in a roughly human way (has eyes, language, linear writing) would probably understand that it encodes meaning, even if it ends up like Linear A and undecipherable.

[Houshalter]

That's a very inefficient way to store the data. You couldn't fit all of wikipedia onto one of those disks like that. I would only do that for the instructions, and pictures would be better than glyphs. Or pictures next to words, so they have at least an idea on how to decrypt it.

Once you introduce a few words, the rest may be decipherable from context, especially with such a large corpus. E.g. certain words will cluster together often, and once you know one, you can guess at the others, which lets you guess at others, and so on.