#NLP 概览与 2026 技术趋势：从规则匹配到大语言模型

from transformers import AutoTokenizer, AutoModelForSequenceClassification
from transformers import pipeline
import torch

def advanced_sentiment_analysis():
    # 加载模型和分词器
    model_name = "uer/roberta-base-finetuned-dianping-chinese"
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModelForSequenceClassification.from_pretrained(model_name)
    
    # 构建情感分析管道（自动处理分词、推理、结果映射）
    classifier = pipeline(
        "sentiment-analysis",
        model=model,
        tokenizer=tokenizer,
        device=0 if torch.cuda.is_available() else -1  # 有GPU用GPU
    )
    
    # 测试文本
    test_texts = [
        "这家火锅的毛肚脆嫩，鸭肠新鲜，下次还要来！",
        "快递太慢了，包装也破了，商品质量一般般",
        "平平无奇的一部电影，没什么亮点也没什么槽点"
    ]
    
    # 输出结果
    for text, res in zip(test_texts, classifier(test_texts)):
        # 把标签转换成中文（原模型标签是正面/负面）
        print(f"📝 文本：{text}")
        print(f"💭 情感：{res['label']}，置信度：{res['score']:.2f}\n")

# 运行
if __name__ == "__main__":
    advanced_sentiment_analysis()

import jieba
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LogisticRegression
from sklearn.pipeline import Pipeline

# 中文预处理：分词
def chinese_tokenize(text):
    return " ".join(jieba.cut(text))

def traditional_sentiment_analysis():
    # 示例训练数据（实际项目中需要更多数据）
    train_data = [
        ("食物很好吃，服务也棒", "正面"),
        ("环境干净，价格合理", "正面"),
        ("非常推荐这家店", "正面"),
        ("快递太慢，包装破损", "负面"),
        ("服务态度恶劣", "负面"),
        ("完全不值这个价格", "负面")
    ]
    train_texts = [chinese_tokenize(t) for t, l in train_data]
    train_labels = [l for t, l in train_data]
    
    # 构建管道：分词预处理 → TF-IDF特征 → 逻辑回归
    pipeline = Pipeline([
        ("tfidf", TfidfVectorizer()),
        ("clf", LogisticRegression())
    ])
    
    # 训练模型
    pipeline.fit(train_texts, train_labels)
    
    # 测试文本
    test_texts = ["毛肚很脆，值得再来", "质量太差了，再也不买了"]
    test_tokens = [chinese_tokenize(t) for t in test_texts]
    
    # 输出结果
    predictions = pipeline.predict(test_tokens)
    probs = pipeline.predict_proba(test_tokens)
    for text, pred, prob in zip(test_texts, predictions, probs):
        print(f"📝 文本：{text}")
        print(f"💭 情感：{pred}，置信度：{max(prob):.2f}\n")

# 运行
if __name__ == "__main__":
    traditional_sentiment_analysis()