多路召回实现

moiveRe/moiveReApp/admin.py

@@ -1,8 +1,9 @@
 from django.contrib import admin
 
 # Register your models here.
-from .models import Question,Rating,Tag
+from .models import Question,Rating,Tag,UserProfile
 
 admin.site.register(Question)
 admin.site.register(Rating)
-admin.site.register(Tag)
+admin.site.register(Tag)
+admin.site.register(UserProfile)

moiveRe/moiveReApp/models.py

@@ -7,6 +7,7 @@
 from django.utils import timezone
 from django.core.validators import MinValueValidator, MaxValueValidator
 from django.contrib.auth.models import User
+from datetime import datetime, timedelta
 
 class Tag(models.Model):
     name = models.CharField(max_length=200)
@@ -27,6 +28,7 @@ class Question(models.Model):
 
     likes = models.PositiveIntegerField(default=0)
     liked_by = models.ManyToManyField(User, blank=True, related_name='liked_questions')
+    clicks = models.PositiveIntegerField(default=0)
     def __str__(self):
         return f"{self.question_text} - {self.img.url}"
 
@@ -60,4 +62,31 @@ def __str__(self):
         return f"{self.question.question_text} - {self.value}"
 
 
+class Userinterests(models.Model):
+    name = models.CharField(max_length=100, unique=True)
+
+    def __str__(self):
+        return self.name
+
+
+class UserProfile(models.Model):
+    # 与用户关联的一对一字段，这里假设你使用了Django内置的User模型
+    user = models.OneToOneField(User, on_delete=models.CASCADE)
+    interests = models.ManyToManyField(Userinterests, related_name='interested_users')
+    liked_items = models.ManyToManyField(Question, blank=True, related_name='liked_users')
+    click_items = models.ManyToManyField(Question, blank=True, related_name='clicked_users')
+
+
+    def __str__(self):
+        return self.user.username + "'s Profile"
+
+    @classmethod
+    def get_user_profile(cls, user):
+        # 根据用户获取或创建用户资料
+        user_profile, created = cls.objects.get_or_create(user=user)
+        return user_profile
+
+
+
+
 

moiveRe/moiveReApp/views.py

@@ -3,7 +3,7 @@
 from .forms import RatingForm
 from django.db.models import Avg
 from django.shortcuts import get_object_or_404
-from .models import Question, Rating
+from .models import Question, Rating ,UserProfile,Userinterests
 import random
 from django.contrib.auth.decorators import login_required
 #from recsystem.userCFrecommendation import usercf
@@ -13,6 +13,7 @@
 import numpy as np
 from sklearn.metrics.pairwise import cosine_similarity
 import math
+from collections import Counter
 
 class IndexView(generic.ListView):
     template_name = "moiveReApp/index.html"
@@ -262,11 +263,43 @@ def generate_recommendations(target_user, item_similarity_matrix, user_item_matr
             # 返回排序后的物品ID列表
             return [item_id for item_id, _ in recommendations]
 # ...
+
+#兴趣标签更新
+def update_uesrinterests(user_profile):
+    # 获取用户喜欢和点击的物品类别
+    liked_categories = [question.category for question in user_profile.liked_items.all()]
+    clicked_categories = [question.category for question in user_profile.click_items.all()]
+    all_categories = liked_categories + clicked_categories
+    # 计算频率
+    category_counter = Counter(all_categories)
+
+    # 根据一定的规则选择兴趣标签;这里我暂时就设置阈值
+    threshold = 4
+    interests = []
+    for category, count in category_counter.items():
+        # 这里可以根据你的规则选择兴趣标签，比如出现频率大于阈值的类别
+        if count >= threshold:
+            interests.append(category)
+
+    # 将选定的兴趣标签添加到用户的兴趣标签列表中
+    #user_profile.interests.clear()  # 清除旧标签保持时效性
+    #print(interests)
+
+    for interest in interests:
+        user_interest, created = Userinterests.objects.get_or_create(name=interest)
+        user_profile.interests.add(user_interest)
+
+
+
 def question_detail(request, question_id):
+    user_profile = UserProfile.get_user_profile(request.user)
     question = get_object_or_404(Question, pk=question_id)
     ratings = Rating.objects.filter(question=question)
     avg_rating = ratings.aggregate(Avg('value'))['value__avg']
     form = RatingForm()
+    user_profile.click_items.add(question)
+    question.clicks+=1
+    update_uesrinterests(user_profile)
 
     if request.method == 'POST':
         form = RatingForm(request.POST)
@@ -318,17 +351,23 @@ def random_questions(request):
 
 @login_required
 def like_question(request, question_id):
+    user_profile = UserProfile.get_user_profile(request.user)
+
     question = get_object_or_404(Question, id=question_id)
 
     if request.user in question.liked_by.all():
         # 用户已经点过赞，取消点赞
         question.liked_by.remove(request.user)
+        user_profile.liked_items.remove(question)
+        update_uesrinterests(user_profile)
         if question.likes>0:
            question.likes-=1
     else:
         # 用户还未点赞，添加点赞
         question.liked_by.add(request.user)
+        user_profile.liked_items.add(question)
         question.likes += 1
+        update_uesrinterests(user_profile)
 
     question.save()
 
@@ -410,4 +449,4 @@ def bertcall(request):
     recommended_movies = Question.objects.filter(id__in=recommendations[:8]) #只展示前ki个
 
     # 在模板中渲染推荐结果
-    return render(request, 'moiveReApp/bertcall.html', {'recommended_movies': recommended_movies})
+    return render(request, 'moiveReApp/bertcall.html', {'recommended_movies': recommended_movies})

moiveRe/recommendationrecall.py

@@ -1,5 +1,7 @@
 
-
+import torch
+from transformers import DistilBertTokenizer, DistilBertModel
+from sklearn.metrics.pairwise import cosine_similarity
 import django
 from django.conf import settings
 import os
@@ -14,6 +16,16 @@
 from sklearn.metrics.pairwise import cosine_similarity
 
 
+from moiveReApp.models import UserProfile,Question
+from django.contrib.auth.models import User
+
+
+user = User.objects.get(username='gly233')
+user_profile = UserProfile.get_user_profile(user)
+question_list = Question.objects.all()
+
+
+
 
 class itemcf():
 
@@ -110,20 +122,121 @@ def generate_recommendations(target_user, item_similarity_matrix, user_item_matr
 
             # 根据感兴趣程度排序推荐列表
             recommendations.sort(key=lambda x: x[1], reverse=True)
-            print(recommendations)
 
 
             # 返回排序后的物品ID列表
-            return [item_id for item_id, _ in recommendations]
+            return recommendations
+
+user_item_matrix, item_dict, user_dict = itemcf.build_user_item_matrix(question_list)
+item_similarity_matrix = itemcf.calculate_user_similarity(user_item_matrix)
+r1 = itemcf.generate_recommendations(user.id, item_similarity_matrix, user_item_matrix, user_dict, item_dict, k=30)[:10]
 
 
 class hotitem():
 
 
     def recall_hot_item(question_list,k):
         sorted_question_list = sorted(question_list, key=lambda q: q.likes, reverse=True)
+        recommendations = [question.id for question in sorted_question_list]
+        return recommendations[:k]
+
+r2 = hotitem.recall_hot_item(question_list,10)
+
+class newitem():
+
+    def recall_new_item(question_list,k):
+        sorted_question_list = sorted(question_list, key=lambda q: q.pub_date, reverse=True)
+        recommendations = [question.id for question in sorted_question_list]
+        return recommendations[:k]
+
+r3 = newitem.recall_new_item(question_list,10)
+
+
+class U2TAG2I():
+
+
+
+    def compute_interest_score(user_profile, questions):
+        interest_scores = {}
+        user_interests = set(user_profile.interests.all())
+        user_interests_names = [interest.name for interest in user_interests]
+        #print(user_interests_names)
+        for question in questions:
+            if  question.category in user_interests_names:
+                interest_scores[question.id] = 0.3
+            else:
+                interest_scores[question.id] = 0
+
+        return interest_scores
+
+    def compute_similarity(user_profile, questions):
+        tokenizer = DistilBertTokenizer.from_pretrained('distilbert-base-uncased')
+        model = DistilBertModel.from_pretrained('distilbert-base-uncased')
+        device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+        model.to(device)
+
+        user_interests_str = ' '.join([interest.name for interest in user_profile.interests.all()])
+
+        question_details = [question.detail for question in questions]
+        num_questions = len(question_details)
+        question_ids = [question.id for question in questions]
+        question_score = []
+        #print(question_ids)
+        question_details.append(user_interests_str)
+        texts = question_details
+
+
+
+        inputs = tokenizer(texts, return_tensors='pt', padding=True, truncation=True)
+
+        # 将张量移动到GPU上
+        input_ids = inputs['input_ids'].to(device)
+        attention_mask = inputs['attention_mask'].to(device)
+        # 使用DistilBERT模型进行推理
+        with torch.no_grad():
+            outputs = model(input_ids=input_ids, attention_mask=attention_mask).last_hidden_state
+
+        # 将文本表示向量转移到CPU上
+        text_embeddings = outputs.mean(dim=1).cpu().numpy()
+
+        # 计算相似度
+        for i in range(num_questions):
+                similarity = cosine_similarity([text_embeddings[i]], [text_embeddings[-1]])[0][0]
+                question_score.append(similarity*0.7)
+
+        # 使用 zip() 函数合并两个列表成元组列表
+        combined = list(zip(question_ids, question_score))
+        # 将元组列表转换为字典
+        interest_score = dict(combined)
+
+        return interest_score
+
+    def generate_recommendations(score1,score2,k):
+            result_dict = {}
+            for key in score1:
+                if key in score2:
+                    result_dict[key] = score1[key] + score2[key]
+
+            sorted_items = sorted(result_dict.items(), key=lambda x: x[1], reverse=True)
+            top_k = sorted_items[:k]
+            return top_k
+
+
+
+a=U2TAG2I.compute_interest_score(user_profile,questions=question_list)
+b=U2TAG2I.compute_similarity(user_profile,questions=question_list)
+
+r4 = U2TAG2I.generate_recommendations(a,b,10)
+
+
+
+print(r1)
+print(r2)
+print(r3)
+print(r4)
+
+
 
-        return sorted_question_list[:k]