A novel scheme for data encryption technique [pdf]

This is vulnerable to frequency analysis, though the paper says "nope". If you encode each word separately, like they suggest, you can very easily do frequency analysis based on word frequency. If you don't split up the blocks by words and use any other fixed-width blocking scheme, you can still do frequency analysis on blocks of that width.

The "key" isn't a key because it's sent with the message, and uses pretty much the same method for "encryption", so the same problem applies. Given that there is no "key", chosen-* attacks don't even make sense. In cryptography you always assume the attacker knows the algorithm but not the key. The "key" is completely determined by the plaintext and algorithm.

I doubt this is a serious paper. Probably an assignment for an intro course that somehow got published. I mean, the description of the algorithm involves "Data.txt"...