bert_and_rf.py

import requests
from bs4 import BeautifulSoup
import csv
import jieba
import pickle
from sklearn.svm import SVR
import pandas as pd
from sklearn.model_selection import GridSearchCV

from transformers import BertTokenizer, BertForSequenceClassification
import torch
from sklearn.metrics import mean_squared_error
from sklearn.ensemble import RandomForestRegressor

def get_sentiment_scores_chinese(file_path):
    df = pd.read_csv(file_path)  # 使用pandas读取csv文件
    comments = df["comments"].tolist()   # 获取comments列并转换成列表
    ratings = df["ratings"].tolist()   # 获取ratings列并转换成列表

    # 载入预训练的Bert模型和tokenizer
    tokenizer = BertTokenizer.from_pretrained('bert-base-chinese')
    model = BertForSequenceClassification.from_pretrained('bert-base-chinese', num_labels=2)
    model.eval()   # 设置为评估模式

    scores = []
    with torch.no_grad():   # 禁用梯度计算
        for comment in comments:
            inputs = tokenizer.encode_plus(comment, add_special_tokens=True, max_length=128, padding='max_length', truncation=True, return_tensors="pt")
            # 对评论进行tokenization并转换为PyTorch张量

            outputs = model(inputs['input_ids'], attention_mask=inputs['attention_mask'])
            # 将输入传入模型进行推理

            logits = outputs[0]
            probs = torch.softmax(logits, dim=-1)[0]   # 计算情感分类得分

            # 将情感分类得分转换为情感分数
            score = float(probs[1].detach().cpu().numpy())   # 取第2个分类标签的概率作为情感分数
            scores.append(score)

    return comments, ratings, scores




c,r,s = get_sentiment_scores_chinese('newdata.csv')


#

def train_regression_model_rf(comments, ratings, scores):
    X = []
    for i in range(len(comments)):
        X.append([scores[i],])
    y = ratings

    # 设置参数范围
    param_grid = {'n_estimators': [50, 100, 200],
                  'max_depth': [10, 20, None]}

    # 创建Random Forest模型
    rf = RandomForestRegressor()

    # 使用GridSearchCV进行交叉验证
    model = GridSearchCV(rf, param_grid, cv=5)
    model.fit(X, y)

    return model.best_estimator_


def predict_rating(model, file_path):
    # Load csv file
    df = pd.read_csv(file_path)
    comments = df["comments"].tolist()
    ratings = df["ratings"].tolist()

    # Use BERT for sentiment analysis and SVR for rating prediction
    tokenizer = BertTokenizer.from_pretrained('bert-base-chinese')
    model_bert = BertForSequenceClassification.from_pretrained('bert-base-chinese', num_labels=2)
    sentiment_scores = []
    for comment in comments:
        encoded_comment = tokenizer(comment, return_tensors='pt', padding=True, truncation=True)
        output = model_bert(**encoded_comment)
        score = output.logits.detach().numpy()[0][1]
        sentiment_scores.append(score)
    predicted_ratings = model.predict([[score] for score in sentiment_scores])

    # Calculate mean squared error
    mse = mean_squared_error(ratings, predicted_ratings)

    # Output predicted ratings and mean squared error
    print(f"Mean squared error: {mse:.2f}")
    print("Predicted ratings:")
    for i, rating in enumerate(predicted_ratings):
        print(f"{i + 1}. {rating:.1f}")


test=train_regression_model_rf(c,r,s) #得到已经训练好的模型
print(test) #打印已经训练好的模型的参数
predict_rating(test, 'new_test.csv')