import java.util.*;

// 用戶評(píng)分項(xiàng)
class Rating {
    int userId;
    int itemId;
    double rating;

    public Rating(int userId, int itemId, double rating) {
        this.userId = userId;
        this.itemId = itemId;
        this.rating = rating;
    }
}

// 用戶評(píng)分?jǐn)?shù)據(jù)存儲(chǔ)
class UserRatings {
    Map<Integer, Double> ratings = new HashMap<>();

    public void addRating(int itemId, double rating) {
        ratings.put(itemId, rating);
    }

    public double getRating(int itemId) {
        return ratings.getOrDefault(itemId, 0.0);
    }

    // 省略其他方法...
}

// 數(shù)據(jù)存儲(chǔ)類（簡(jiǎn)化的示例）
class DataModel {
    Map<Integer, UserRatings> userRatings = new HashMap<>();

    public void addRating(Rating rating) {
        UserRatings userRating = userRatings.getOrDefault(rating.userId, new UserRatings());
        userRating.addRating(rating.itemId, rating.rating);
        userRatings.put(rating.userId, userRating);
    }

    // 省略其他方法...
}

public class UserBasedCollaborativeFiltering {

    // 計(jì)算兩個(gè)用戶之間的皮爾遜相關(guān)系數(shù)
    public static double pearsonCorrelation(UserRatings user1, UserRatings user2) {
        // ... 這里省略了計(jì)算皮爾遜相關(guān)系數(shù)的具體實(shí)現(xiàn) ...
        // 需要考慮共同評(píng)分項(xiàng)、評(píng)分均值等
        return 0.0; // 示例返回0，實(shí)際應(yīng)計(jì)算皮爾遜相關(guān)系數(shù)
    }

    // 為指定用戶生成推薦
    public static List<Integer> generateRecommendations(DataModel dataModel, int targetUserId, int numRecommendations) {
        List<Integer> recommendations = new ArrayList<>();

        // 獲取目標(biāo)用戶的評(píng)分?jǐn)?shù)據(jù)
        UserRatings targetUserRatings = dataModel.userRatings.get(targetUserId);
        if (targetUserRatings == null) {
            return recommendations; // 如果沒有評(píng)分?jǐn)?shù)據(jù)，則無(wú)法推薦
        }

        // 計(jì)算目標(biāo)用戶與其他用戶的相似度
        Map<Integer, Double> userSimilarities = new HashMap<>();
        for (int userId : dataModel.userRatings.keySet()) {
            if (userId != targetUserId) {
                UserRatings otherUserRatings = dataModel.userRatings.get(userId);
                double similarity = pearsonCorrelation(targetUserRatings, otherUserRatings);
                if (similarity > 0) { // 只考慮相似度大于0的用戶
                    userSimilarities.put(userId, similarity);
                }
            }
        }

        // 根據(jù)相似用戶的評(píng)分生成推薦
        // ... 這里省略了根據(jù)相似度生成推薦的具體實(shí)現(xiàn) ...
        // 需要遍歷相似用戶評(píng)過(guò)分的項(xiàng)目，并考慮相似度進(jìn)行排序

        // 示例：隨機(jī)添加一些推薦ID
        for (int i = 0; i < numRecommendations && i < 100; i++) { // 假設(shè)最多推薦100首
            recommendations.add(i + 1000); // 示例ID，實(shí)際應(yīng)用中應(yīng)基于相似用戶評(píng)分計(jì)算
        }

        return recommendations;
    }

    // 主函數(shù)示例
    public static void main(String[] args) {
        DataModel dataModel = new DataModel();
        // ... 加載或添加評(píng)分?jǐn)?shù)據(jù)到dataModel中 ...

        // 為用戶ID為1的用戶生成5個(gè)推薦
        List<Integer> recommendations = generateRecommendations(dataModel, 1, 5);
        for (int itemId : recommendations) {
            System.out.println("推薦ID: " + itemId);
        }
    }
}

Map<String, Map<String, Double>> userRatings = new HashMap<>();
// 假設(shè)我們有兩個(gè)用戶和三個(gè)物品
userRatings.put("user1", Map.of("item1", 5.0, "item2", 3.0, "item3", 4.0));
userRatings.put("user2", Map.of("item1", 4.0, "item2", 4.5, "item3", 2.0));
// ... 為其他用戶添加評(píng)分

public double cosineSimilarity(Map<String, Double> ratings1, Map<String, Double> ratings2) {
    double dotProduct = 0.0;
    double norm1 = 0.0;
    double norm2 = 0.0;

    // 找出兩個(gè)用戶共同評(píng)分的物品
    Set<String> commonItems = new HashSet<>(ratings1.keySet());
    commonItems.retainAll(ratings2.keySet());

    for (String item : commonItems) {
        dotProduct += ratings1.get(item) * ratings2.get(item);
        norm1 += Math.pow(ratings1.get(item), 2);
        norm2 += Math.pow(ratings2.get(item), 2);
    }

    if (norm1 == 0 || norm2 == 0) {
        return 0.0; // 如果沒有共同評(píng)分的物品，返回0
    }

    double similarity = dotProduct / (Math.sqrt(norm1) * Math.sqrt(norm2));
    return similarity;
}