Python打造在线食谱推荐服务

在这个快节奏的时代，如何快速找到心仪的食谱成为许多美食爱好者的难题。今天，我们将用Python语言构建一个在线食谱推荐服务，帮助用户轻松找到心仪的美食。

Python打造在线食谱推荐服务

项目背景

随着互联网的发展，食谱资源越来越丰富，但用户往往需要在海量信息中筛选，耗时费力。我们的目标是通过Python编程，结合机器学习算法，为用户提供个性化的食谱推荐。

技术选型

前端展示：使用Flask框架构建Web界面。
后端处理：利用Scikit-learn进行食谱数据的分类和推荐。
数据存储：使用SQLite数据库存储食谱信息。

项目步骤

1. 数据收集与预处理

首先，我们需要收集大量的食谱数据。可以通过爬虫技术从各大美食网站获取数据，并将其存储在SQLite数据库中。

import sqlite3
import requests
from bs4 import BeautifulSoup

def fetch_recipes(url):
    response = requests.get(url)
    soup = BeautifulSoup(response.content, 'html.parser')
    recipes = []
    for item in soup.find_all('div', class_='recipe'):
        title = item.find('h3').text
        ingredients = item.find('ul', class_='ingredients').text
        recipes.append((title, ingredients))
    return recipes

def save_to_db(recipes):
    conn = sqlite3.connect('recipes.db')
    c = conn.cursor()
    c.execute('''CREATE TABLE IF NOT EXISTS recipes (title TEXT, ingredients TEXT)''')
    c.executemany('INSERT INTO recipes VALUES (?, ?)', recipes)
    conn.commit()
    conn.close()

url = 'https://example-recipe-website.com'
recipes = fetch_recipes(url)
save_to_db(recipes)

2. 构建Web界面

使用Flask框架构建一个简单的Web界面，用户可以通过输入关键词来搜索食谱。

from flask import Flask, request, render_template

app = Flask(__name__)

@app.route('/')
def index():
    return render_template('index.html')

@app.route('/search', methods=['POST'])
def search():
    keyword = request.form['keyword']
    conn = sqlite3.connect('recipes.db')
    c = conn.cursor()
    c.execute('SELECT * FROM recipes WHERE title LIKE ?', ('%' + keyword + '%',))
    results = c.fetchall()
    conn.close()
    return render_template('results.html', results=results)

if __name__ == '__main__':
    app.run(debug=True)

3. 个性化推荐算法

利用Scikit-learn的TF-IDF和KNN算法，根据用户的搜索历史和偏好，推荐相似的食谱。

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.neighbors import NearestNeighbors

def recommend_recipes(keyword, n=5):
    conn = sqlite3.connect('recipes.db')
    c = conn.cursor()
    c.execute('SELECT ingredients FROM recipes')
    ingredients_list = [row[0] for row in c.fetchall()]
    conn.close()

    vectorizer = TfidfVectorizer()
    tfidf_matrix = vectorizer.fit_transform(ingredients_list)

    model = NearestNeighbors(n_neighbors=n, metric='cosine')
    model.fit(tfidf_matrix)

    keyword_vector = vectorizer.transform([keyword])
    distances, indices = model.kneighbors(keyword_vector)

    recommended_recipes = []
    for idx in indices[0]:
        recommended_recipes.append(ingredients_list[idx])
    return recommended_recipes

@app.route('/recommend', methods=['POST'])
def recommend():
    keyword = request.form['keyword']
    recommendations = recommend_recipes(keyword)
    return render_template('recommendations.html', recommendations=recommendations)