- It seems, magnetization plateau is called a entangled state, if so, how ?
That is at best an imprecise statement. As discussed in the "Magnetization Plateaus" chapter by Takigawa, M. and Mila, F. in the book Lacroix, C., Mendels, P., Mila, F. (eds) Introduction to Frustrated Magnetism, Springer (2010) there are different mechanisms that can generate magnetization plateaus. The most well-studied is related to order-by-disorder, in which a long-ranged classical state is stabilized by thermal or quantum fluctuations. This would be a product state, without entanglement. However, the literature also points out "quantum plateaus", in which quantum states are stabilized. These can be entangled.
Note that you can have both unentangled and entangled plateaus within the same system; see for example Nishimoto, S., Shibata, N. and Hotta, C., Nat. Commun. 4, 2287 (2013) discussing the kagome Heisenberg antiferromagnet. Their numerical results indicate that the $1/9$ plateau is a state with finite topological entanglement entropy. However, the nature of this state is still under debate; see Fang, D-z., Xi, N., Ran, S-J. and Su, G. Phys. Rev. B 107, L220401 (2023) arguing that the entanglement entropy in the state follows the critical scaling of a $c=1$ conformal field theory. Either scenario indicates a state with nontrivial entanglement properties.
- What could be the physical application of phenomena of magnetization plateau ?
I'm not aware of any specific proposed technological applications. In principle, robust plateaus may be exploitable in future devices (c.f. quantum Hall effects). For now, I would consider the phenomenon largely of basic scientific interest. If specific plateaus can realize interesting topological phases, perhaps these would be more robust to chemical disorder than zero-field realizations of the same phases.
- How observance of plateau is associated with breaking of some symmetry in the spin system (I think it is transnational).
Magnetization plateaus can be observed in systems both with and without translational symmetry breaking. For example, the $1/3$ plateau in triangular lattices breaks the translational symmetry of the crystal as its magnetic unit cell involves three sites rather than one, but the $1/2$ plateau in bond-alternating chains does not as both the magnetic and conventional unit cells involve two sites. See table 10.1 in the book chapter linked above for more examples.