An unrooted tree looks like a messy spiderweb. But if you pick up the spiderweb from one node, it dangles cleanly with that node at the top. That node becomes the root.
Rooting a tree creates a hierarchy. Every node except the root has exactly one parent. Nodes can have zero or more children. This parent-child structure is what makes tree algorithms work.
The same tree can be rooted at different nodes, giving you different hierarchies. The choice of root depends on the problem you are solving.
