There are two parts to the GIS issue. Once created, you have systems/software through which you are able to view the maps, query them for information and may use to analyze them for further information. A typical example is to find the shortest distance between two points. Since a machine is involved you could throw in additional data before the analysis. Variables such as traffic patterns, snow conditions, construction and other information can be used to arrive at an optimum route.
There are several software programs available from well known vendors such as ESRI, Autodesk and others. A common use is in those handy car GPS navigators that we use. We get our position mapped on such Intelligent/electronic maps. It is the GIS mapping that give you the information. The actual GPS data just gives your current position on the map.
The second part of the issue is creation of maps in a format that can be used by the software to get you meaningful data. One way of creating such maps would be to actually do a survey with accurate co-ordinates read from GPS devices. You also can convert accurate satellite imagery into maps. A third approach is to use existing paper maps. Once these are scanned you get an image which represents every point on the picture as a dot. However, you cannot attach any information to it and it is heavy in terms of storage needs.
All you really need for the computer to understand a line are the two endpoints and their co-ordinates. Software is used with those two points to create a line. Similarly you convert all the other types of elements too. After the information is then attached, these become the POI or the points of interest in your GPS navigation maps.