当前位置：首页 > news >正文

用Python做有趣的AI项目 2：用 Python 和 NLTK 构建一个聊天机器人

news 来源：原创 2025/4/28 6:47:48

✅ 项目目标

创建一个基础聊天机器人，能够通过简单规则或意图匹配理解用户的输入，并返回合适的回复。这个项目会让你了解 NLP（自然语言处理）的基础流程。

🛠️ 所需环境和依赖

安装必要库：

bash
pip install nltk numpy
首次使用 NLTK 还需要下载语言包（只需要一次）：

python
import nltk
nltk.download('punkt')        # 分词器
nltk.download('wordnet')      # 词根还原
nltk.download('omw-1.4')      # WordNet 数据集

🧠 聊天逻辑简介

我们将采用一种经典方式：意图分类（intent classification），用户的输入会匹配某个“意图”，然后返回相应的回答。

📁 第一步：准备聊天意图文件（intents.json）
先建一个 JSON 文件（命名为 intents.json）：

json
{"intents": [{"tag": "问候","patterns": ["你好", "嗨", "在吗", "您好", "哈喽"],"responses": ["你好！有什么我可以帮您的吗？", "嗨，很高兴见到你！"]},{"tag": "再见","patterns": ["再见", "拜拜", "回见", "下次见"],"responses": ["再见啦，祝你愉快！", "拜拜！期待下次聊天。"]},{"tag": "感谢","patterns": ["谢谢", "感谢你", "多谢", "太好了"],"responses": ["不客气！", "随时为您服务 :)"]},{"tag": "天气","patterns": ["今天天气怎么样", "天气", "气温如何"],"responses": ["我不是天气预报员，但你可以看看手机天气哦~"]}]
}

🧱 第二步：构建数据预处理模块

python
import json
import random
import numpy as np
import nltk
from nltk.stem import WordNetLemmatizer

加载数据

with open('intents.json', 'r', encoding='utf-8') as f:intents = json.load(f)lemmatizer = WordNetLemmatizer()

分析意图数据

words = []
classes = []
documents = []for intent in intents['intents']:for pattern in intent['patterns']:word_list = nltk.word_tokenize(pattern)words.extend(word_list)documents.append((word_list, intent['tag']))if intent['tag'] not in classes:classes.append(intent['tag'])

词干化、去重

words = [lemmatizer.lemmatize(w.lower()) for w in words if w.isalnum()]
words = sorted(set(words))
classes = sorted(set(classes))

🧠 第三步：构建训练数据

pythontraining = []
output_empty = [0] * len(classes)for doc in documents:bag = []word_patterns = [lemmatizer.lemmatize(w.lower()) for w in doc[0]]for w in words:bag.append(1 if w in word_patterns else 0)output_row = list(output_empty)output_row[classes.index(doc[1])] = 1training.append([bag, output_row])

转成 NumPy 数组

random.shuffle(training)
training = np.array(training, dtype=object)train_x = np.array(list(training[:, 0]))
train_y = np.array(list(training[:, 1]))

🧠 第四步：训练一个简单的神经网络模型

pythonfrom tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropoutmodel = Sequential()
model.add(Dense(128, input_shape=(len(train_x[0]),), activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(64, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(len(train_y[0]), activation='softmax'))model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])model.fit(train_x, train_y, epochs=200, batch_size=5, verbose=1)
model.save('chatbot_model.h5')

🗨️ 第五步：实现聊天功能

pythonimport random
from tensorflow.keras.models import load_modelmodel = load_model('chatbot_model.h5')def clean_up_sentence(sentence):sentence_words = nltk.word_tokenize(sentence)sentence_words = [lemmatizer.lemmatize(word.lower()) for word in sentence_words]return sentence_wordsdef bag_of_words(sentence, words):sentence_words = clean_up_sentence(sentence)bag = [0] * len(words)for s in sentence_words:for i, w in enumerate(words):if w == s:bag[i] = 1return np.array(bag)def classify(sentence):bow = bag_of_words(sentence, words)res = model.predict(np.array([bow]))[0]thresh = 0.25results = [(i, r) for i, r in enumerate(res) if r > thresh]results.sort(key=lambda x: x[1], reverse=True)return classes[results[0][0]] if results else "无匹配"def get_response(intent_tag):for intent in intents['intents']:if intent['tag'] == intent_tag:return random.choice(intent['responses'])

聊天循环

print("你好，我是智能助手（输入 '退出' 来结束对话）")
while True:message = input("你：")if message.lower() in ['退出', 'bye', 'exit']:print("机器人：再见啦！")breakintent = classify(message)response = get_response(intent)print("机器人：", response)