initial
Browse files- .gitignore +18 -0
- Dockerfile +14 -0
- README copy.md +0 -0
- recommend_normal.py +75 -0
- recommendwithdesc.py +61 -0
- recommendwithhist.py +151 -0
- requirements.txt +0 -0
- vercel.json +15 -0
.gitignore
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
.idea/
|
2 |
+
.vscode/
|
3 |
+
.venv*/
|
4 |
+
venv*/
|
5 |
+
__pycache__/
|
6 |
+
dist/
|
7 |
+
.coverage*
|
8 |
+
htmlcov/
|
9 |
+
.tox/
|
10 |
+
docs/_build/
|
11 |
+
*.pkl
|
12 |
+
*.pyc
|
13 |
+
__pycache__/
|
14 |
+
venv/
|
15 |
+
.env
|
16 |
+
instance/
|
17 |
+
*.sqlite3
|
18 |
+
.vercel
|
Dockerfile
ADDED
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Use the official Python 3.10.9 image
|
2 |
+
FROM python:3.10.9
|
3 |
+
|
4 |
+
# Copy the current directory contents into the container at .
|
5 |
+
COPY . .
|
6 |
+
|
7 |
+
# Set the working directory to /
|
8 |
+
WORKDIR /
|
9 |
+
|
10 |
+
# Install requirements.txt
|
11 |
+
RUN pip install --no-cache-dir --upgrade -r /requirements.txt
|
12 |
+
|
13 |
+
# Start the FastAPI app on port 7860, the default port expected by Spaces
|
14 |
+
CMD ["python", "app.py"]
|
README copy.md
ADDED
Binary file (178 Bytes). View file
|
|
recommend_normal.py
ADDED
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import pickle
|
2 |
+
from sklearn.metrics.pairwise import cosine_similarity
|
3 |
+
import numpy as np
|
4 |
+
import pandas as pd
|
5 |
+
import os
|
6 |
+
import difflib
|
7 |
+
|
8 |
+
from huggingface_hub import hf_hub_download
|
9 |
+
|
10 |
+
repo_id = "Navanihk/recommendationsystemmovie"
|
11 |
+
def load_data():
|
12 |
+
try:
|
13 |
+
|
14 |
+
# Download the CSV file
|
15 |
+
csv_path = hf_hub_download(repo_id=repo_id, filename="movieswithposter_updated.csv")
|
16 |
+
|
17 |
+
# Load as DataFrame
|
18 |
+
movies_data = pd.read_csv(csv_path)
|
19 |
+
return movies_data
|
20 |
+
except Exception as e:
|
21 |
+
print(f"Error loading data from Hugging Face: {e}")
|
22 |
+
# Fallback to local file if available
|
23 |
+
if os.path.exists('./movieswithposter_updated.csv'):
|
24 |
+
return pd.read_csv('./movieswithposter_updated.csv')
|
25 |
+
else:
|
26 |
+
raise
|
27 |
+
|
28 |
+
# Load movie data
|
29 |
+
movies_data = load_data()
|
30 |
+
vectorizer_path = hf_hub_download(repo_id=repo_id, filename="feature_vector.pkl")
|
31 |
+
similarity_path = hf_hub_download(repo_id=repo_id, filename="model_similarity.pkl")
|
32 |
+
def recommend_movies(movie_name):
|
33 |
+
# Add the movie to the user's history
|
34 |
+
if vectorizer_path and similarity_path:
|
35 |
+
# Load the vectorizer and similarity matrix
|
36 |
+
with open(vectorizer_path, 'rb') as vec_file, open(similarity_path, 'rb') as sim_file:
|
37 |
+
vectorizer = pickle.load(vec_file)
|
38 |
+
similarity = pickle.load(sim_file)
|
39 |
+
|
40 |
+
print(f"Movies suggested for you based on your past choices: \n")
|
41 |
+
|
42 |
+
# Create an aggregate similarity score across all movies in history
|
43 |
+
combined_similarity = np.zeros(similarity.shape[0])
|
44 |
+
|
45 |
+
for past_movie in [movie_name]:
|
46 |
+
# Find a close match for each movie in the user's history
|
47 |
+
list_of_all_titles = movies_data['title'].tolist()
|
48 |
+
find_close_match = difflib.get_close_matches(past_movie, list_of_all_titles)
|
49 |
+
|
50 |
+
if find_close_match:
|
51 |
+
close_match = find_close_match[0]
|
52 |
+
# Find the index of the movie in the dataset
|
53 |
+
index_of_the_movie = movies_data[movies_data.title == close_match]['index'].values[0]
|
54 |
+
# Accumulate the similarity scores
|
55 |
+
combined_similarity += similarity[index_of_the_movie]
|
56 |
+
|
57 |
+
# Sort movies based on the combined similarity score
|
58 |
+
sorted_similar_movies = list(enumerate(combined_similarity))
|
59 |
+
sorted_similar_movies = sorted(sorted_similar_movies, key=lambda x: x[1], reverse=True)
|
60 |
+
|
61 |
+
# Recommend the top movies that the user hasn't already seen
|
62 |
+
i = 1
|
63 |
+
movie_returns = []
|
64 |
+
for movie in sorted_similar_movies:
|
65 |
+
index = movie[0]
|
66 |
+
# title_from_index = movies_data[movies_data.index == index]['title'].values[0]
|
67 |
+
dataFromtitle = movies_data[movies_data.index == index]
|
68 |
+
movie_returns.append({'title':dataFromtitle['title'].values[0],'image':dataFromtitle['poster'].values[0]})
|
69 |
+
print(i, '.',dataFromtitle['title'].values[0], "(Score:", round(movie[1], 2), ")")
|
70 |
+
|
71 |
+
i+=1
|
72 |
+
if i > 35: # Limit recommendations to top 5
|
73 |
+
break
|
74 |
+
return movie_returns
|
75 |
+
|
recommendwithdesc.py
ADDED
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import pickle
|
2 |
+
from sklearn.metrics.pairwise import cosine_similarity
|
3 |
+
import numpy as np
|
4 |
+
import pandas as pd
|
5 |
+
|
6 |
+
from huggingface_hub import hf_hub_download
|
7 |
+
|
8 |
+
repo_id = "Navanihk/recommendationsystemmovie"
|
9 |
+
def load_data():
|
10 |
+
try:
|
11 |
+
|
12 |
+
# Download the CSV file
|
13 |
+
csv_path = hf_hub_download(repo_id=repo_id, filename="movieswithposter_updated.csv")
|
14 |
+
|
15 |
+
# Load as DataFrame
|
16 |
+
movies_data = pd.read_csv(csv_path)
|
17 |
+
return movies_data
|
18 |
+
except Exception as e:
|
19 |
+
print(f"Error loading data from Hugging Face: {e}")
|
20 |
+
# Fallback to local file if available
|
21 |
+
if os.path.exists('./movieswithposter_updated.csv'):
|
22 |
+
return pd.read_csv('./movieswithposter_updated.csv')
|
23 |
+
else:
|
24 |
+
raise
|
25 |
+
|
26 |
+
# Load movie data
|
27 |
+
movies_data = load_data()
|
28 |
+
model_vectorizer = hf_hub_download(repo_id=repo_id, filename="model_vectorizer.pkl")
|
29 |
+
similarity_path = hf_hub_download(repo_id=repo_id, filename="model_similarity.pkl")
|
30 |
+
with open(model_vectorizer, 'rb') as vec_file, open(similarity_path, 'rb') as sim_file:
|
31 |
+
vectorizer = pickle.load(vec_file)
|
32 |
+
similarity = pickle.load(sim_file)
|
33 |
+
def recommend_movies_with_desc(query):
|
34 |
+
# Transform the query into a feature vector using the same vectorizer
|
35 |
+
feature_vecto = vectorizer.transform(query)
|
36 |
+
with open('feature_vector.pkl', 'rb') as feature:
|
37 |
+
feature_vectors = pickle.load(feature)
|
38 |
+
|
39 |
+
# Calculate cosine similarity between the query vector and the feature vectors of the movies
|
40 |
+
sim = cosine_similarity(feature_vectors, feature_vecto)
|
41 |
+
|
42 |
+
# Extract the similarity scores for the query against all movies
|
43 |
+
combined_similarity = sim.flatten()
|
44 |
+
|
45 |
+
# Sort the movies by similarity score
|
46 |
+
sorted_similar_movies = list(enumerate(combined_similarity))
|
47 |
+
sorted_similar_movies = sorted(sorted_similar_movies, key=lambda x: x[1], reverse=True)
|
48 |
+
|
49 |
+
# Print out the top 5 similar movies
|
50 |
+
i = 1
|
51 |
+
movie_recom=[]
|
52 |
+
for movie in sorted_similar_movies:
|
53 |
+
index = movie[0]
|
54 |
+
# title_from_index = movies_data.iloc[index]['title'] # Assuming movies_data is a DataFrame
|
55 |
+
dataFromtitle = movies_data[movies_data.index == index]
|
56 |
+
movie_recom.append({'title':dataFromtitle['title'].values[0],'image':dataFromtitle['poster'].values[0]})
|
57 |
+
print(i, '.',dataFromtitle['title'].values[0], "(Score:", round(movie[1], 2), ")")
|
58 |
+
i += 1
|
59 |
+
if i > 35: # Limit recommendations to top 5
|
60 |
+
break
|
61 |
+
return movie_recom
|
recommendwithhist.py
ADDED
@@ -0,0 +1,151 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import numpy as np
|
2 |
+
import pandas as pd
|
3 |
+
import difflib
|
4 |
+
from sklearn.feature_extraction.text import TfidfVectorizer
|
5 |
+
from sklearn.metrics.pairwise import cosine_similarity
|
6 |
+
import pickle
|
7 |
+
import os
|
8 |
+
|
9 |
+
import nltk
|
10 |
+
nltk.download('punkt')
|
11 |
+
|
12 |
+
from huggingface_hub import hf_hub_download
|
13 |
+
nltk.download('punkt_tab')
|
14 |
+
from nltk.stem import PorterStemmer
|
15 |
+
from nltk.tokenize import word_tokenize
|
16 |
+
repo_id = "Navanihk/recommendationsystemmovie"
|
17 |
+
def stemmed_tokenizer(text):
|
18 |
+
ps = PorterStemmer()
|
19 |
+
words = word_tokenize(text)
|
20 |
+
return [ps.stem(word) for word in words]
|
21 |
+
# Initialize an empty dictionary to store user history
|
22 |
+
user_history = {}
|
23 |
+
# Function to save user history to a pickle file
|
24 |
+
def save_user_history():
|
25 |
+
with open('user_history.pkl', 'wb') as file:
|
26 |
+
pickle.dump(user_history, file)
|
27 |
+
|
28 |
+
# Function to load user history from a pickle file
|
29 |
+
def load_user_history():
|
30 |
+
global user_history
|
31 |
+
if os.path.exists('user_history.pkl'):
|
32 |
+
with open('user_history.pkl', 'rb') as file:
|
33 |
+
user_history = pickle.load(file)
|
34 |
+
|
35 |
+
# Load movie data
|
36 |
+
# movies_data = pd.read_csv('./movieswithposter_updated.csv')
|
37 |
+
def load_data():
|
38 |
+
try:
|
39 |
+
|
40 |
+
# Download the CSV file
|
41 |
+
csv_path = hf_hub_download(repo_id=repo_id, filename="movieswithposter_updated.csv")
|
42 |
+
|
43 |
+
# Load as DataFrame
|
44 |
+
movies_data = pd.read_csv(csv_path)
|
45 |
+
return movies_data
|
46 |
+
except Exception as e:
|
47 |
+
print(f"Error loading data from Hugging Face: {e}")
|
48 |
+
# Fallback to local file if available
|
49 |
+
if os.path.exists('./movieswithposter_updated.csv'):
|
50 |
+
return pd.read_csv('./movieswithposter_updated.csv')
|
51 |
+
else:
|
52 |
+
raise
|
53 |
+
|
54 |
+
# Load movie data
|
55 |
+
movies_data = load_data()
|
56 |
+
|
57 |
+
# Pre-process data
|
58 |
+
selected_features = ['genres', 'keywords', 'tagline', 'cast', 'director']
|
59 |
+
for feature in selected_features:
|
60 |
+
movies_data[feature] = movies_data[feature].fillna('')
|
61 |
+
|
62 |
+
# Combine features
|
63 |
+
combined_features = movies_data['genres'] + ' ' + movies_data['keywords'] + ' ' + movies_data['tagline'] + ' ' + movies_data['cast'] + ' ' + movies_data['director']
|
64 |
+
model_vectorizer = hf_hub_download(repo_id=repo_id, filename="model_vectorizer.pkl")
|
65 |
+
similarity_path = hf_hub_download(repo_id=repo_id, filename="model_similarity.pkl")
|
66 |
+
# Check if the model (vectorizer and similarity) exists
|
67 |
+
if model_vectorizer and similarity_path:
|
68 |
+
# Load the vectorizer and similarity matrix
|
69 |
+
with open(model_vectorizer, 'rb') as vec_file, open(similarity_path, 'rb') as sim_file:
|
70 |
+
vectorizer = pickle.load(vec_file)
|
71 |
+
similarity = pickle.load(sim_file)
|
72 |
+
else:
|
73 |
+
# Train the model if it doesn't exist
|
74 |
+
vectorizer = TfidfVectorizer(stop_words='english',tokenizer=stemmed_tokenizer)
|
75 |
+
feature_vectors = vectorizer.fit_transform(combined_features)
|
76 |
+
with open('feature_vector.pkl', 'wb') as file:
|
77 |
+
pickle.dump(feature_vectors, file)
|
78 |
+
# Calculate cosine similarity
|
79 |
+
similarity = cosine_similarity(feature_vectors)
|
80 |
+
|
81 |
+
# Save the model (vectorizer and similarity matrix)
|
82 |
+
with open('model_vectorizer.pkl', 'wb') as vec_file, open('model_similarity.pkl', 'wb') as sim_file:
|
83 |
+
pickle.dump(vectorizer, vec_file)
|
84 |
+
pickle.dump(similarity, sim_file)
|
85 |
+
|
86 |
+
# Function to recommend movies based on both user input and history
|
87 |
+
def recommend_movieswithhistory(user_id, movie_name):
|
88 |
+
# Add the movie to the user's history
|
89 |
+
add_to_history(user_id, movie_name)
|
90 |
+
print(user_id,movie_name)
|
91 |
+
# Fetch the user's history
|
92 |
+
history = get_history(user_id)
|
93 |
+
|
94 |
+
if len(history) == 0:
|
95 |
+
print("No history found for the user.")
|
96 |
+
return
|
97 |
+
|
98 |
+
print(f"Movies suggested for you based on your past choices: {history}\n")
|
99 |
+
|
100 |
+
# Create an aggregate similarity score across all movies in history
|
101 |
+
combined_similarity = np.zeros(similarity.shape[0])
|
102 |
+
|
103 |
+
for past_movie in history:
|
104 |
+
# Find a close match for each movie in the user's history
|
105 |
+
list_of_all_titles = movies_data['title'].tolist()
|
106 |
+
find_close_match = difflib.get_close_matches(past_movie, list_of_all_titles)
|
107 |
+
|
108 |
+
if find_close_match:
|
109 |
+
close_match = find_close_match[0]
|
110 |
+
# Find the index of the movie in the dataset
|
111 |
+
index_of_the_movie = movies_data[movies_data.title == close_match]['index'].values[0]
|
112 |
+
# Accumulate the similarity scores
|
113 |
+
combined_similarity += similarity[index_of_the_movie]
|
114 |
+
|
115 |
+
# Sort movies based on the combined similarity score
|
116 |
+
sorted_similar_movies = list(enumerate(combined_similarity))
|
117 |
+
sorted_similar_movies = sorted(sorted_similar_movies, key=lambda x: x[1], reverse=True)
|
118 |
+
|
119 |
+
# Recommend the top movies that the user hasn't already seen
|
120 |
+
i = 1
|
121 |
+
movie_return=[]
|
122 |
+
for movie in sorted_similar_movies:
|
123 |
+
index = movie[0]
|
124 |
+
# title_from_index = movies_data[movies_data.index == index]['title'].values[0]
|
125 |
+
dataFromtitle = movies_data[movies_data.index == index]
|
126 |
+
|
127 |
+
|
128 |
+
if dataFromtitle['title'].values[0] not in history: # Don't recommend movies the user has already interacted with
|
129 |
+
|
130 |
+
print(i, '.',dataFromtitle['title'].values[0], "(Score:", round(movie[1], 2), ")")
|
131 |
+
movie_return.append({'title':dataFromtitle['title'].values[0],'image':dataFromtitle['poster'].values[0]})
|
132 |
+
i += 1
|
133 |
+
if i > 35: # Limit recommendations to top 5
|
134 |
+
break
|
135 |
+
return movie_return
|
136 |
+
|
137 |
+
# Function to add a movie to user history
|
138 |
+
def add_to_history(user_id, movie_title):
|
139 |
+
if user_id not in user_history:
|
140 |
+
user_history[user_id] = []
|
141 |
+
user_history[user_id].append(movie_title)
|
142 |
+
save_user_history() # Save the updated history after adding a movie
|
143 |
+
|
144 |
+
# Function to get movies from user history
|
145 |
+
def get_history(user_id):
|
146 |
+
return user_history.get(user_id, [])
|
147 |
+
|
148 |
+
# Load the user history at the start of the program
|
149 |
+
load_user_history()
|
150 |
+
|
151 |
+
|
requirements.txt
ADDED
Binary file (1.32 kB). View file
|
|
vercel.json
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"version": 2,
|
3 |
+
"builds": [
|
4 |
+
{
|
5 |
+
"src": "./main.py",
|
6 |
+
"use": "@vercel/python"
|
7 |
+
}
|
8 |
+
],
|
9 |
+
"routes": [
|
10 |
+
{
|
11 |
+
"src": "/(.*)",
|
12 |
+
"dest": "/"
|
13 |
+
}
|
14 |
+
]
|
15 |
+
}
|