1
def __init__(self, num_topics=2, passes=10, random_state=42):
2
    self.num_topics = num_topics  # 主题数量（如2个主题）
3
    self.passes = passes          # 训练迭代次数（如10次）
4
    self.random_state = random_state  # 随机种子（确保结果可复现）
5
    self.dictionary = None        # 词典（词汇到ID的映射）
6
    self.corpus = None            # 语料库（文档的词频统计）
7
    self.lda_model = None         # LDA模型对象
8
    self.documents = []          # 文档列表（原始文本）

1
def load_corpus(self, file_path):
2
    """加载语料数据"""
3
    try:
4
        with open(file_path, 'r', encoding='utf-8') as f:
5
            self.documents = [line.strip().split() for line in f if line.strip()]
6
        logging.info(f"成功加载 {len(self.documents)} 篇文档")
7
        return self.documents
8
    except FileNotFoundError:
9
        logging.error(f"文件不存在: {file_path}")
10
        return []

1
def build_dictionary_and_corpus(self):
2
    """构建词典和语料库"""
3
    if not self.documents:
4
        logging.error("没有可用的文档数据")
5
        return None, None
6

7
    # 创建词典，词典：词汇→ID映射
8
    self.dictionary = corpora.Dictionary(self.documents)
9
    logging.info(f"词典大小: {len(self.dictionary)}")
10

11
    # 构建语料库，语料库：文档→词频统计
12
    self.corpus = [self.dictionary.doc2bow(doc) for doc in self.documents]
13
    return self.dictionary, self.corpus

1
def train_model(self):
2
    """训练LDA模型"""
3
    if self.corpus is None:
4
        self.build_dictionary_and_corpus()
5

6
    self.lda_model = models.LdaModel(
7
        corpus=self.corpus,
8
        id2word=self.dictionary,
9
        num_topics=self.num_topics,
10
        passes=self.passes,
11
        random_state=self.random_state
12
    )
13
    return self.lda_model

1
def evaluate_model(self):
2
    """评估模型性能"""
3
    if self.lda_model is None:
4
        self.train_model()
5

6
    coherence_model = CoherenceModel(
7
        model=self.lda_model,
8
        texts=self.documents,
9
        dictionary=self.dictionary,
10
        coherence='c_v'
11
    )
12
    coherence_score = coherence_model.get_coherence()
13
    return coherence_score

1
# 使用pyLDAvis生成可视化
2
vis_data = gensimvis.prepare(self.lda_model, self.corpus, self.dictionary)  # 生成可视化数据
3

4
# 保存可视化为HTML文件（可选）
5
pyLDAvis.save_html(vis_data, "lda_visualization.html")  # 保存到当前目录
6

7
# 直接展示可视化（可选，需支持HTML渲染的环境）
8
pyLDAvis.display(vis_data)  # 在Jupyter Notebook等支持HTML的环境中直接展示