A dipole can be understood by thinking of it as two charges, q and -q, held apart from each other by a small distance d. So a dipole is neutral overall but it will respond to electric fields since it wants to align with field lines, and it will accelerate in regions of nonconstant field, since (in the q, -q picture) different magnitude forces will act on the two sides of the dipole.
An "ideal dipole" would correspond to the limit d→0, qd=constant.
A dipole moment qd characterizes the strength of a dipole. The larger the value qd, the more torque the dipole will feel in a uniform field, or alternatively, the more force it will feel in a field gradient.
There are other ways to understand dipoles too. For example, all magnets are dipoles (there is no such thing as a magnetic "charge" or monopole). But one could think of a magnetic dipole as a small conducting loop where an ideal magnetic dipole corresponds to the limit of the loop's diameter approaching zero. (This is a more realistic model than thinking of the magnetic dipole as two oppositely charged magnetic monopoles.)
Localized charge configurations can be analyzed in the "far field" regime (r>>d) in terms of their "monopole moment", corresponding to the net charge of the distribution, their "dipole moment", corresponding to the distribution's net dipole moment, their "quadrupole moment" (a quadrupole can be thought of as two opposing dipoles separated by d), their "octopole moment" (two opposing quadrupoles separated by d), etc. If you specify all the 2n-pole moments and their direction, this completely specifies the field generated by any localized charge distribution, provided you're interested in the field much farther away from the distribution than the distribution's size.