The whole point of assembler is that there is a direct, 1:1 correspondence between the assembler mnemonics and the processor opcodes - so you always know precisely what code will be generated for a particular assembler source line.

As soon as you start abstracting to a higher level - eg, with algebraic notation - you remove this certainty, since there will very often be more than one possible way to implement an algebraic expression in the underlying instruction set; eg, there are several ways to implement the algebraic expression a = 0

So, if you value programmer convenience over deterministic code generation, why not just use a high-level language like 'C'?!