In modern social network applications, recommendation systems have become an essential function. Whether it is recommending friends to users, recommending topics of interest, recommending related products, or recommending more valuable content, recommendation systems can effectively improve user experience and stickiness.
In this article, we will introduce how to use Java to write a social network application based on a recommendation system. We will combine actual code and detailed steps to help readers quickly understand and implement a basic recommendation system.
1. Data collection and processing
Before implementing any recommendation system, we need to collect and process a large amount of data. In social network applications, user information, posts, comments, likes and other data are all valuable sources of data.
To facilitate demonstration, we can use an open source virtual data generator to generate these data. The specific steps are as follows:
2. Representation of users and items
In the recommendation system, we need to convert users and items into vector or matrix form in order to calculate their similarity or Make recommendations. In social network applications, we can use the following methods to represent users and items:
User A = [1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 1, 0, 0 , 0, 1, 0, 0, 1]
The vector length is 24, and each position represents a topic or post. 1 means that user A has followed the topic or published the post, and 0 means not.
Post A = [1, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 10 , 0]
The vector length is 24, and each position represents a label or statistical data. 1 means the post contains the tag or content, 0 means it does not.
3. User-based collaborative filtering recommendation
User-based collaborative filtering is a common method in recommendation systems. It recommends items based on the similarity of user interests. Here, we use user-based collaborative filtering to recommend suitable posts for users. The specific steps are as follows:
The following is the Java code implementation of the algorithm:
public class CollaborativeFiltering {
/**
* 计算用户间的皮尔逊相关系数
* @param user1 用户1
* @param user2 用户2
* @param data 数据集
* @return 皮尔逊相关系数
*/
public double pearsonCorrelation(Map<Integer, Double> user1, Map<Integer, Double> user2,
Map<Integer, Map<Integer, Double>> data) {
double sum1 = 0, sum2 = 0, sum1Sq = 0, sum2Sq = 0, pSum = 0;
int n = 0;
for (int item : user1.keySet()) {
if (user2.containsKey(item)) {
sum1 += user1.get(item);
sum2 += user2.get(item);
sum1Sq += Math.pow(user1.get(item), 2);
sum2Sq += Math.pow(user2.get(item), 2);
pSum += user1.get(item) * user2.get(item);
n++;
}
}
if (n == 0)
return 0;
double num = pSum - (sum1 * sum2 / n);
double den = Math.sqrt((sum1Sq - Math.pow(sum1, 2) / n) *
(sum2Sq - Math.pow(sum2, 2) / n));
if (den == 0)
return 0;
return num / den;
}
/**
* 基于用户的协同过滤推荐算法
* @param data 数据集
* @param userId 目标用户 ID
* @param K 最相似的 K 个用户
* @param N 推荐的 N 个帖子
* @return 推荐的帖子 ID 列表
*/
public List<Integer> userBasedCollaborativeFiltering(Map<Integer, Map<Integer, Double>> data,
int userId, int K, int N) {
Map<Integer, Double> targetUser = data.get(userId); // 目标用户
List<Map.Entry<Integer, Double>> similarUsers = new ArrayList<>(); // 与目标用户兴趣相似的用户
for (Map.Entry<Integer, Map<Integer, Double>> entry: data.entrySet()) {
int id = entry.getKey();
if (id == userId)
continue;
double sim = pearsonCorrelation(targetUser, entry.getValue(), data); // 计算皮尔逊相关系数
if (sim > 0)
similarUsers.add(new AbstractMap.SimpleEntry<>(id, sim));
}
Collections.sort(similarUsers, (a, b) -> b.getValue().compareTo(a.getValue())); // 按相似度从高到低排序
List<Integer> itemIds = new ArrayList<>();
for (int i = 0; i < K && i < similarUsers.size(); i++) {
Map.Entry<Integer, Double> entry = similarUsers.get(i);
int userId2 = entry.getKey();
Map<Integer, Double> user2 = data.get(userId2);
for (int itemId: user2.keySet()) {
if (!targetUser.containsKey(itemId)) { // 如果目标用户没看过该帖子
itemIds.add(itemId);
}
}
}
Map<Integer, Double> scores = new HashMap<>();
for (int itemId: itemIds) {
double score = 0;
int count = 0;
for (Map.Entry<Integer, Double> entry: similarUsers) {
int userId2 = entry.getKey();
Map<Integer, Double> user2 = data.get(userId2);
if (user2.containsKey(itemId)) { // 如果用户 2 看过该帖子
score += entry.getValue() * user2.get(itemId);
count++;
if (count == N)
break;
}
}
scores.put(itemId, score);
}
List<Integer> pickedItemIds = new ArrayList<>();
scores.entrySet().stream().sorted((a, b) -> b.getValue().compareTo(a.getValue()))
.limit(N).forEach(entry -> pickedItemIds.add(entry.getKey())); // 按得分从高到低排序并选出前N个
return pickedItemIds;
}
}4. Content-based recommendation algorithm
The content-based recommendation algorithm is another type of recommendation system. A common method that recommends items based on the similarity of their attributes. Here, we use a content-based recommendation algorithm to recommend suitable posts to users. The specific steps are as follows:
The following is the Java code implementation of the content-based recommendation algorithm:
public class ContentBasedRecommendation {
/**
* 计算两个向量的余弦相似度
* @param v1 向量1
* @param v2 向量2
* @return 余弦相似度
*/
public double cosineSimilarity(double[] v1, double[] v2) {
double dotProduct = 0;
double norma = 0;
double normb = 0;
for (int i = 0; i < v1.length; i++) {
dotProduct += v1[i] * v2[i];
norma += Math.pow(v1[i], 2);
normb += Math.pow(v2[i], 2);
}
if (norma == 0 || normb == 0)
return 0;
return dotProduct / (Math.sqrt(norma) * Math.sqrt(normb));
}
/**
* 基于内容的推荐算法
* @param data 数据集
* @param userId 目标用户 ID
* @param N 推荐的 N 个帖子
* @return 推荐的帖子 ID 列表
*/
public List<Integer> contentBasedRecommendation(Map<Integer, Map<Integer, Double>> data,
int userId, int N) {
Map<Integer, Double> targetUser = data.get(userId); // 目标用户
int[] pickedItems = new int[data.size()];
double[][] itemFeatures = new double[pickedItems.length][24]; // 物品特征矩阵
for (Map.Entry<Integer, Map<Integer, Double>> entry: data.entrySet()) {
int itemId = entry.getKey();
Map<Integer, Double> item = entry.getValue();
double[] feature = new double[24];
for (int i = 0; i < feature.length; i++) {
if (item.containsKey(i+1)) {
feature[i] = item.get(i+1);
} else {
feature[i] = 0;
}
}
itemFeatures[itemId-1] = feature; // 物品 ID 从 1 开始，需要减一
}
for (int itemId: targetUser.keySet()) {
pickedItems[itemId-1] = 1; // 物品 ID 从 1 开始，需要减一
}
double[] similarities = new double[pickedItems.length];
for (int i = 0; i < similarities.length; i++) {
if (pickedItems[i] == 0) {
similarities[i] = cosineSimilarity(targetUser.values().stream().mapToDouble(Double::doubleValue).toArray(), itemFeatures[i]);
}
}
List<Integer> itemIds = new ArrayList<>();
while (itemIds.size() < N) {
int maxIndex = -1;
for (int i = 0; i < similarities.length; i++) {
if (pickedItems[i] == 0 && (maxIndex == -1 || similarities[i] > similarities[maxIndex])) {
maxIndex = i;
}
}
if (maxIndex == -1 || similarities[maxIndex] < 0) {
break; // 找不到更多相似的物品了
}
itemIds.add(maxIndex + 1); // 物品 ID 从 1 开始，需要加一
pickedItems[maxIndex] = 1;
}
Map<Integer, Double> scores = new HashMap<>();
for (int itemId: itemIds) {
double[] features = itemFeatures[itemId-1]; // 物品 ID 从 1 开始，需要减一
double score = cosineSimilarity(targetUser.values().stream().mapToDouble(Double::doubleValue).toArray(), features);
scores.put(itemId, score);
}
List<Integer> pickedItemIds = new ArrayList<>();
scores.entrySet().stream().sorted((a, b) -> b.getValue().compareTo(a.getValue()))
.limit(N).forEach(entry -> pickedItemIds.add(entry.getKey())); // 按得分从高到低排序并选出前N个
return pickedItemIds;
}
}5. Integrate the recommendation algorithm into the application
After completing the above two recommendation algorithms Once implemented, we can integrate them into the application. The specific steps are as follows:
The following is the Java code implementation of the application:
@RestController
@RequestMapping("/recommendation")
public class RecommendationController {
private CollaborativeFiltering collaborativeFiltering = new CollaborativeFiltering();
private ContentBasedRecommendation contentBasedRecommendation = new ContentBasedRecommendation();
@Autowired
private UserService userService;
@GetMapping("/userbased/{userId}")
public List<Integer> userBasedRecommendation(@PathVariable Integer userId) {
List<User> allUsers = userService.getAllUsers();
Map<Integer, Map<Integer, Double>> data = new HashMap<>();
for (User user: allUsers) {
Map<Integer, Double> userVector = new HashMap<>();
List<Topic> followedTopics = user.getFollowedTopics();
for (Topic topic: followedTopics) {
userVector.put(topic.getId(), 1.0);
}
List<Post> posts = user.getPosts();
for (Post post: posts) {
userVector.put(post.getId() + 1000, 1.0);
}
List<Comment> comments = user.getComments();
for (Comment comment: comments) {
userVector.put(comment.getId() + 2000, 1.0);
}
List<Like> likes = user.getLikes();
for (Like like: likes) {
userVector.put(like.getId() + 3000, 1.0);
}
data.put(user.getId(), userVector);
}
List<Integer> itemIds = collaborativeFiltering.userBasedCollaborativeFiltering(data, userId, 5, 10);
return itemIds;
}
@GetMapping("/contentbased/{userId}")
public List<Integer> contentBasedRecommendation(@PathVariable Integer userId) {
List<User> allUsers = userService.getAllUsers();
Map<Integer, Map<Integer, Double>> data = new HashMap<>();
for (User user: allUsers) {
Map<Integer, Double> userVector = new HashMap<>();
List<Topic> followedTopics = user.getFollowedTopics();
for (Topic topic: followedTopics) {
userVector.put(topic.getId(), 1.0);
}
List<Post> posts = user.getPosts();
for (Post post: posts) {
userVector.put(post.getId() + 1000, 1.0);
}
List<Comment> comments = user.getComments();
for (Comment comment: comments) {
userVector.put(comment.getId() + 2000, 1.0);
}
List<Like> likes = user.getLikes();
for (Like like: likes) {
userVector.put(like.getId() + 3000, 1.0);
}The above is the detailed content of How to write a recommendation system-based social network application using Java. For more information, please follow other related articles on the PHP Chinese website!
![[Web front-end] Node.js quick start](https://img.php.cn/upload/course/000/000/067/662b5d34ba7c0227.png)
