使用 NLP 和 SQLite 进行餐厅评论分析

通常情况下，很多企业由于没有利润，缺乏适当的改进措施而以失败告终。大多数情况下，餐馆老板在提高生产力方面面临很多困难。该项目确实帮助了那些想要提高生产力的人，从而增加了他们的业务利润。这是该项目的主要目标。

该项目的作用是，餐厅老板通过 ML 分类算法（朴素贝叶斯）处理的客户文本评论了解他餐厅的缺点，例如他餐厅最不喜欢的食物项目，并将其结果存储在数据库中使用SQLite。

使用的工具和技术：

• NLTK
• 机器学习
• Python
• 特金特
• Sqlite3
• 熊猫

分步实施：

步骤 1：导入库和初始化数据

首先，我们导入 NumPy、matplotlib、pandas、nltk、re、sklearn、Tkinter、sqlite3 库，用于数据操作、文本数据处理、模式识别、数据训练、图形用户界面和数据库数据操作。

Python3

import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
import re
import nltk
from nltk.corpus import stopwords
from nltk.stem.porter import PorterStemmer
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.model_selection import train_test_split
from sklearn.naive_bayes import GaussianNB
from sklearn.metrics import accuracy_score
from tkinter import *
import sqlite3
  
dataset = pd.read_csv('Restaurant_Reviews.tsv',
                      delimiter='\t', quoting=3)
corpus = []
rras_code = "Wyd^H3R"
food_rev = {}
food_perc = {}
  
conn = sqlite3.connect('Restaurant_food_data.db')
c = conn.cursor()
  
for i in range(0, 1000):
    review = re.sub('[^a-zA-Z]', ' ', dataset['Review'][i])
    review = review.lower()
    review = review.split()
  
    ps = PorterStemmer()
    all_stopwords = stopwords.words('english')
    all_stopwords.remove('not')
  
    review = [ps.stem(word)
              for word in review if not word in set(all_stopwords)]
      
    review = ' '.join(review)
    corpus.append(review)
  
cv = CountVectorizer(max_features=1500)
X = cv.fit_transform(corpus).toarray()
y = dataset.iloc[:, -1].values
  
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.20, random_state=0)
  
classifier = GaussianNB()
classifier.fit(X_train, y_train)
  
y_pred = classifier.predict(X_test)
  
variables = []
clr_variables = []
  
foods = ["Idly", "Dosa", "Vada", "Roti", "Meals", "Veg Biryani",
         "Egg Biryani", "Chicken Biryani", "Mutton Biryani",
         "Ice Cream", "Noodles", "Manchooriya", "Orange juice",
         "Apple Juice", "Pineapple juice", "Banana juice"]
  
for i in foods:
    food_rev[i] = []
    food_perc[i] = [0.0, 0.0]
  
  
def init_data():
  
    conn = sqlite3.connect('Restaurant_food_data.db')
    c = conn.cursor()
    for i in range(len(foods)):
        c.execute("INSERT INTO item VALUES(:item_name,:no_of_customers,\
        :no_of_positives,:no_of_negatives,:pos_perc,:neg_perc)",
                  {
                      'item_name': foods[i],
                      'no_of_customers': "0",
                      'no_of_positives': "0",
                      'no_of_negatives': "0",
                      'pos_perc': "0.0%",
                      'neg_perc': "0.0%"
                  }
                  )
    conn.commit()
    conn.close()

第 2 步：澄清用户

最初，我们的 GUI 应用程序会询问用户他是所有者还是客户，以决定要执行的操作。

蟒蛇3

root1 = Tk()
main = "Restaurant Review Analysis System/"
root1.title(main+"Welcome Page")
  
label = Label(root1, text="RESTAURANT REVIEW ANALYSIS SYSTEM",
              bd=2, font=('Arial', 47, 'bold', 'underline'))
  
ques = Label(root1, text="Are you a Customer or Owner ???")
  
cust = Button(root1, text="Customer", font=('Arial', 20),
              padx=80, pady=20, command=take_review)
  
owner = Button(root1, text="Owner", font=('Arial', 20),
               padx=100, pady=20, command=login)
  
'''conn=sqlite3.connect('Restaurant_food_data.db')
c=conn.cursor()
c.execute("CREATE TABLE item (Item_name text,No_of_customers text,\
No_of_positive_reviews text,No_of_negative_reviews text,Positive_percentage \
text,Negative_percentage text) ")
conn.commit()
conn.close()'''
#c.execute("DELETE FROM item")
  
root1.attributes("-zoomed", True)
label.grid(row=0, column=0)
ques.grid(row=1, column=0, sticky=W+E)
ques.config(font=("Helvetica", 30))
cust.grid(row=2, column=0)
owner.grid(row=3, column=0)
conn.commit()
conn.close()
root1.mainloop()

澄清用户

第 3 步：收集数据

一旦系统确保用户是客户，它就会以文本格式要求食品评论。顾客必须选择他从餐厅拿走的食品，然后对选定的食品进行评论。当他点击提交按钮时，文本评论会在 ML 算法下进行处理，以预测它是正面评论还是负面评论。然后，将整个数据插入到数据库中。

蟒蛇3

def take_review():
    root2 = Toplevel()
    root2.title(main+"give review")
  
    label = Label(root2, text="RESTAURANT REVIEW ANALYSIS SYSTEM",
                  bd=2, font=('Arial', 47, 'bold', 'underline'))
  
    req1 = Label(root2, text="Select the item(s) you have taken.....")
  
    conn = sqlite3.connect('Restaurant_food_data.db')
    c = conn.cursor()
  
    chk_btns = []
    selected_foods = []
    req2 = Label(root2, text="Give your review below....")
    rev_tf = Entry(root2, width=125, borderwidth=5)
    req3 = Label(root2, text="NOTE : Use not instead of n't.")
    global variables
    variables = []
    chk_btns = []
  
    for i in range(len(foods)):
        var = IntVar()
        chk = Checkbutton(root2, text=foods[i], variable=var)
        variables.append(var)
        chk_btns.append(chk)
  
    label.grid(row=0, column=0, columnspan=4)
    req1.grid(row=1, column=0, columnspan=4, sticky=W+E)
    req1.config(font=("Helvetica", 30))
  
    for i in range(4):
        for j in range(4):
            c = chk_btns[i*4+j]
            c.grid(row=i+3, column=j, columnspan=1, sticky=W)
    selected_foods = []
  
    submit_review = Button(root2, text="Submit Review", font=(
        'Arial', 20), padx=100, pady=20, command=lambda: [
        estimate(rev_tf.get()), root2.destroy()])
  
    root2.attributes("-zoomed", True)
    req2.grid(row=7, column=0, columnspan=4, sticky=W+E)
    req2.config(font=("Helvetica", 20))
    rev_tf.grid(row=8, column=1, rowspan=3, columnspan=2, sticky=S)
    req3.grid(row=11, column=1, columnspan=2)
    submit_review.grid(row=12, column=0, columnspan=4)
    conn.commit()
    conn.close()
  
# Processing and storing the data
def estimate(s):
  
    conn = sqlite3.connect('Restaurant_food_data.db')
    c = conn.cursor()
    review = re.sub('[^a-zA-Z]', ' ', s)
    review = review.lower()
    review = review.split()
    ps = PorterStemmer()
    all_stopwords = stopwords.words('english')
    all_stopwords.remove('not')
      
    review = [ps.stem(word)
              for word in review if not word in set(all_stopwords)]
    review = ' '.join(review)
    X = cv.transform([review]).toarray()
    res = classifier.predict(X)  # list
      
    if "not" in review:
        res[0] = abs(res[0]-1)
      
    selected_foods = []
    for i in range(len(foods)):
        if variables[i].get() == 1:
            selected_foods.append(foods[i])
      
    c.execute("SELECT *,oid FROM item")
    records = c.fetchall()
      
    for i in records:
        rec = list(i)
        if rec[0] in selected_foods:
            n_cust = int(rec[1])+1
            n_pos = int(rec[2])
            n_neg = int(rec[3])
              
            if res[0] == 1:
                n_pos += 1
            else:
                n_neg += 1
                  
            pos_percent = round((n_pos/n_cust)*100, 1)
            neg_percent = round((n_neg/n_cust)*100, 1)
            c.execute("""UPDATE item SET Item_name=:item_name,No_of_customers\
            =:no_of_customers,No_of_positive_reviews=:no_of_positives,\
            No_of_negative_reviews=:no_of_negatives,Positive_percentage\
            =:pos_perc,Negative_percentage=:neg_perc  where oid=:Oid""",
                      {
                          'item_name': rec[0],
                          'no_of_customers': str(n_cust),
                          'no_of_positives': str(n_pos),
                          'no_of_negatives': str(n_neg),
                          'pos_perc': str(pos_percent)+"%",
                          'neg_perc': str(neg_percent)+"%",
                          'Oid': foods.index(rec[0])+1
                      }
                      )
    selected_foods = []
  
    conn.commit()
    conn.close()

接受审查

第 4 步：验证所有权

如果系统的当前用户是餐厅的老板，那么系统会通过询问 rras_code（即全球唯一标识餐厅的代码。而且是高度机密的，不应共享此代码）来验证所有者与其他人（餐厅的共同所有者除外）。

蟒蛇3

def login():
    root3 = Toplevel()
    root3.title(main+"owner verfication")
      
    label = Label(root3, text="RESTAURANT REVIEW ANALYSIS SYSTEM",
                  bd=2, font=('Arial', 47, 'bold', 'underline'))
      
    label2 = Label(root3, text="VERIFY OWNERSHIP", bd=1,
                   font=('Helvetica', 30, 'bold', 'underline'))
      
    label3 = Label(root3, text="To verify your ownership, please \
    enter your restaurant's private rras code....",
                   bd=1, font=('Helvetica', 20, 'bold'))
    ent = Entry(root3, show="*", borderwidth=2)
    submit_code = Button(root3, text="Submit", font=('Arial', 20), padx=80,
                         pady=20, command=lambda: [
                           view_details(ent.get()), root3.destroy()])
      
    root3.attributes("-zoomed", True)
    label.grid(row=0, column=0, columnspan=3)
    label2.grid(row=1, column=0, sticky=W+E, columnspan=3)
    label3.grid(row=2, column=0, sticky=W, columnspan=3)
    ent.grid(row=3, column=1, columnspan=1)
    submit_code.grid(row=4, column=1, columnspan=1)

验证所有权

第 5 步：访问数据

所有权验证后，所有者在新页面上有 3 个选项，如下所述：

蟒蛇3

def popup():
    messagebox.showerror("Error Message!", "Incorrect code!")
  
  
def view_details(s):
    if(s != rras_code):
        popup()
    else:
        root4 = Toplevel()
        root4.title(main+"view_details")
        label = Label(root4, text="RESTAURANT REVIEW ANALYSIS SYSTEM",
                      bd=2, font=('Arial', 47, 'bold', 'underline'))
        sug1 = Label(
            root4, text="Click the below button, if you want to view\
            the data from your database....")
          
        acc_btn = Button(root4, text="View Data", font=(
            'Arial', 20), padx=100, pady=20, command=access_data)
          
        sug2 = Label(root4, text="Click the below button, if you want \
        to clear specific item data...")
          
        itemclr_btn = Button(root4, text="Clear Item Data", font=(
            'Arial', 20), padx=100, pady=20, command=clr_itemdata)
          
        sug3 = Label(root4, text="Click the below button, if you want to\
        clear all item data...")
          
        allclr_btn = Button(root4, text="Clear All Data", font=(
            'Arial', 20), padx=100, pady=20, command=clr_alldata)
          
        exit_btn = Button(root4, text="Exit", command=root4.destroy)
  
        root4.attributes("-zoomed", True)
        label.grid(row=0, column=0)
        sug1.grid(row=1, column=0)
        sug1.config(font=("Helvetica", 30))
        acc_btn.grid(row=2, column=0)
        sug2.grid(row=3, column=0)
        sug2.config(font=("Helvetica", 30))
        itemclr_btn.grid(row=4, column=0)
        sug3.grid(row=5, column=0)
        sug3.config(font=("Helvetica", 30))
        allclr_btn.grid(row=6, column=0)
        exit_btn.grid(row=9, column=0, sticky=S)

选项

步骤 6：查看数据

所有者可以查看数据库中的数据，其中每种食物具有诸如客户数量、正面评价数量、负面评价数量、正面率、负面率等属性。评价高的食品用绿色标记，评价最差的食品用红色标记，以便于理解数据摘要。现在，店主可以识别低评价的食品，并尝试通过一些措施来改善食品的口味，例如请新厨师等，这肯定会改善他的生意。

蟒蛇3

def access_data():
    root5 = Toplevel()
    root5.title(main+"Restaurant_Database")
      
    label = Label(root5, text="RESTAURANT REVIEW ANALYSIS SYSTEM",
                  bd=2, font=('Arial', 47, 'bold', 'underline'))
      
    title1 = Label(root5, text="S.NO", font=('Arial', 10, 'bold', 'underline'))
      
    title2 = Label(root5, text="FOOD ITEM", font=(
        'Arial', 10, 'bold', 'underline'))
      
    title3 = Label(root5, text="NO.OF CUSTOMERS",
                   font=('Arial', 10, 'bold', 'underline'))
      
    title4 = Label(root5, text="NO.OF POSITIVE REVIEWS",
                   font=('Arial', 10, 'bold', 'underline'))
      
    title5 = Label(root5, text="NO.OF NEGATIVE REVIEWS",
                   font=('Arial', 10, 'bold', 'underline'))
      
    title6 = Label(root5, text="POSITIVE RATE",
                   font=('Arial', 10, 'bold', 'underline'))
      
    title7 = Label(root5, text="NEGATIVE RATE",
                   font=('Arial', 10, 'bold', 'underline'))
  
    label.grid(row=0, column=0, columnspan=7)
    title1.grid(row=1, column=0)
    title2.grid(row=1, column=1)
    title3.grid(row=1, column=2)
    title4.grid(row=1, column=3)
    title5.grid(row=1, column=4)
    title6.grid(row=1, column=5)
    title7.grid(row=1, column=6)
  
    conn = sqlite3.connect('Restaurant_food_data.db')
    c = conn.cursor()
  
    c.execute("SELECT *,oid from item")
    records = c.fetchall()
    pos_rates = []
  
    for record in records:
        record = list(record)
        pos_rates.append(float(record[-3][:-1]))
  
    max_pos = max(pos_rates)
    min_pos = min(pos_rates)
  
    for i in range(len(records)):
        rec_list = list(records[i])
        if str(max_pos)+"%" == rec_list[-3]:
            rec_lab = [Label(root5, text=str(rec_list[-1]), fg="green")]
            for item in rec_list[:-1]:
                lab = Label(root5, text=item, fg="green")
                rec_lab.append(lab)
        elif str(min_pos)+"%" == rec_list[-3]:
            rec_lab = [Label(root5, text=str(rec_list[-1]), fg="red")]
            for item in rec_list[:-1]:
                lab = Label(root5, text=item, fg="red")
                rec_lab.append(lab)
        else:
            rec_lab = [Label(root5, text=str(rec_list[-1]))]
            for item in rec_list[:-1]:
                lab = Label(root5, text=item)
                rec_lab.append(lab)
  
        for j in range(len(rec_lab)):
            rec_lab[j].grid(row=i+2, column=j)
  
    exit_btn = Button(root5, text="Exit", command=root5.destroy)
    exit_btn.grid(row=len(records)+5, column=3)
  
    conn.commit()
    conn.close()
    root5.attributes("-zoomed", True)

数据库内容

第七步：清除数据：

当所有者进行了一些调整或修改时，可以清除那些特定的项目数据，以便他可以快速观察食品项目的性能。如果所有者愿意，他也可以清除所有食品项目数据。

蟒蛇3

def clr_itemdata():
    root6 = Toplevel()
    root6.title(main+"clear_item_data")
      
    label = Label(root6, text="RESTAURANT REVIEW ANALYSIS SYSTEM",
                  bd=2, font=('Arial', 47, 'bold', 'underline'))
      
    req1 = Label(root6, text="Pick the items to clear their corresponding\
    item data....")
      
    chk_list = []
    global clr_variables
    clr_variables = []
      
    for i in range(len(foods)):
        var = IntVar()
        chk = Checkbutton(root6, text=foods[i], variable=var)
        clr_variables.append(var)
        chk_list.append(chk)
      
    label.grid(row=0, column=0, columnspan=4)
    req1.grid(row=1, column=0, columnspan=4, sticky=W+E)
    req1.config(font=("Helvetica", 30))
      
    for i in range(4):
        for j in range(4):
            c = chk_list[i*4+j]
            c.grid(row=i+3, column=j, columnspan=1, sticky=W)
      
    clr_item = Button(root6, text="Clear", font=(
        'Arial', 20), padx=100, pady=20, command=lambda: [
      clr_data(), root6.destroy()])
      
    clr_item.grid(row=8, column=0, columnspan=4)
    root6.attributes("-zoomed", True)
  
  
def clr_alldata():
    confirm = messagebox.askquestion(
        "Confirmation", "Are you sure to delete all data??")
      
    if confirm == "yes":
        conn = sqlite3.connect('Restaurant_food_data.db')
        c = conn.cursor()
        for i in range(len(foods)):
            c.execute("""UPDATE item SET Item_name=:item_name,No_of_customers\
            =:no_of_customers,No_of_positive_reviews=:no_of_positives,\
            No_of_negative_reviews=:no_of_negatives,Positive_percentage=:\
            pos_perc,Negative_percentage=:neg_perc  where oid=:Oid""",
                      {
                          'item_name': foods[i],
                          'no_of_customers': "0",
                          'no_of_positives': "0",
                          'no_of_negatives': "0",
                          'pos_perc': "0.0%",
                          'neg_perc': "0.0%",
                          'Oid': i+1
                      }
                      )
        conn.commit()
        conn.close()

清除食品项目数据

最后，这是我通过技术提高企业生产力的想法。这样，业务问题就会因生产力的提高而停止。

现实生活中的项目应用：

• 它可用于任何美食餐厅/酒店。
• 在直接改善业务的食品改进测量中有效。
• 方便使用的。
• 没有商业损失的机会。