func simpleItemCF() {
m := model.NewGenericDataModel()
add := func(userId, itemId uint64, value float64) {
p := model.NewGenericPreference(userId, itemId, value)
m.AddPreference(p)
}
// A用户喜欢物品A,C
// B用户喜欢物品A,B,C
// C用户喜欢物品A
// 将C物品推荐给用户C
add(1, 1, 5.0)
add(1, 3, 4.0)
add(2, 1, 4.0)
add(2, 2, 5.0)
add(2, 3, 5.0)
add(3, 1, 5.0)
s := similarity.NewPearsonCorrelationSimilarity(m)
r := recommender.NewGenericItemBasedRecommender(m, s)
rItems, err := r.Recommend(3, 1)
if err != nil {
log.Fatal(err)
}
for _, item := range rItems {
log.Println("item:", item)
}
}
func main() {
flag.Parse()
itemInfos, err := loadMovieData()
if err != nil {
log.Fatal(err)
}
file, err := os.Open("ratings.dat")
if err != nil {
log.Fatal(err)
}
m, err := model.NewFileDataModel(file, "::")
if err != nil {
log.Fatal(err)
}
s := similarity.NewPearsonCorrelationSimilarity(m)
log.Println("User", *userId, "preferences: ")
userPrefs, _ := m.GetUserPreferences(*userId)
for i, itemId := range userPrefs.Ids() {
itemName, ok := itemInfos[itemId]
if !ok {
continue
}
log.Println(itemId, itemName, userPrefs.Values()[i])
}
if *what == "user" {
log.Println("<UserCF>")
n := recommender.NewNearestNUserNeighborhood(100, -100.0, m, s)
ur := recommender.NewGenericUserBasedRecommender(m, n, s)
rItems, err := ur.Recommend(*userId, *howMany)
if err != nil {
log.Fatal(err)
}
for _, item := range rItems {
itemName, ok := itemInfos[item.ItemId]
if !ok {
continue
}
log.Println("item:", item, itemName)
}
}
if *what == "item" {
log.Println("<ItemCF>")
r := recommender.NewGenericItemBasedRecommender(m, s)
rItems, err := r.Recommend(*userId, *howMany)
if err != nil {
log.Fatal(err)
}
for _, item := range rItems {
itemName, ok := itemInfos[item.ItemId]
if !ok {
continue
}
log.Println("item:", item, itemName)
}
}
}