The explanation is given in frequecy domain......try to undetstand in this way that you have sine wave siganl with some amplititude a.sin(wt) ....now the fourier transform will represent two peaks at w and -w negative peak if you are sampling it you will need alleast 2w frequecy sampling in order to get proper sampling in the other wise you will get alalising in the sampled signal....same explanation is given there.....
Look at this way when you pass a signal through the LTI ( system like filter) in the time domain it dose the convolution operation and in frequecy domain it dose the multiplication operation....from the basic identities of the transform theroy.....Now why the interpolation term is used ..... the term interpolation is used to find out the the data realtion or one way to say co-relation with the the data.....when you are sampling it out there is not a single frequecy component exitis there existis a band of frequecy while sampling ....so it is clubeed iffect of all the frequecy that you need to consider.....That is reason the term interoplation is used
The term is introduced, and in my opinion basically explained in problem text. You have time discrete signal samples and want to design a time-continuous reconstruction of the original signal. Linear interpolation is the simple method of connecting two point with a straight line. You can do it with pencil, paper and a ruler. This is in my view a straightforward and visual method to reconstruct a signal.
The problem does a mathematical analysis of this simple, linear interpolated signal. That's all.