Implementing user ratings

Many online vendors ask their customers to rate the products that they purchase, typically on a scale of one to five stars. We can modify our previous Recommender2 program to incorporate these numerical ratings. To test the new version, we'll also modify our DataGenerator and Filter programs.

The modified DataGenerator program is shown in Listing 9.18:

Listing 9.18: Program to generate random ratings

(The folded code is the same as in the DataGenerator1 program in Listing 9.2.) It creates random ratings from the set {1.0, 1.5, 2.0, 2.5, …, 5.0} that are normally distributed with mean 3.0 and standard deviation 1.0.

The only modifications needed for the Filter program are changing int to double where necessary. The results from a sample run are shown in Figure 9.10 and Figure 9.11:

Figure 9.10: Utility file from DataGenerator3

Figure 9.11: Similarity file from DataGenerator3

Item #9 has not been bought by anyone—all five of its entries in the utility matrix are 0.0. Consequently...