论文摘要生成器：用TextRank算法实现文献关键信息提取

我们基于python代码，使用PyQt5创建图形用户界面（GUI），同时支持中英文两种语言的文本论文文献关键信息提取。

PyQt5：用于创建GUI应用程序。
jieba：中文分词库，用于中文文本的处理。
re：正则表达式模块，用于文本清理和句子分割。
numpy：提供数值计算能力，如数组操作、矩阵运算等，主要用于TextRank算法的实现。

import sys
import re
import jieba
import numpy as np
from PyQt5.QtWidgets import (QApplication, QMainWindow, QWidget, QVBoxLayout,
                             QHBoxLayout, QTextEdit, QPushButton, QLabel,
                             QMessageBox, QSpinBox, QFileDialog, QComboBox)
from PyQt5.QtCore import Qt


class TextRankSummarizer:
    def __init__(self, language='chinese'):
        self.language = language
        self.stopwords = self.load_stopwords()

        # 初始化jieba中文分词器
        if language == 'chinese':
            jieba.initialize()

    def load_stopwords(self):
        """内置停用词表"""
        if self.language == 'chinese':
            return {'的', '了', '在', '是', '我', '有', '和', '就', '不', '人', '都', '一', '一个', '也', '要'}
        else:  # 英文停用词
            return {'a', 'an', 'the', 'and', 'or', 'but', 'if', 'is', 'are', 'of', 'to', 'in', 'on'}

    def preprocess_text(self, text):
        """文本预处理"""
        # 清洗特殊字符
        text = re.sub(r'[^\w\s。，.?！]', '', text)

        # 分句处理
        if self.language == 'chinese':
            sentences = re.split(r'[。！？]', text)
        else:
            sentences = re.split(r'[.!?]', text)

        return [s.strip() for s in sentences if len(s) > 2]

    def calculate_similarity(self, sentence, other_sentence):
        """计算句子相似度"""
        words1 = [w for w in (jieba.cut(sentence) if self.language == 'chinese' else sentence.lower().split())
                  if w not in self.stopwords]
        words2 = [w for w in
                  (jieba.cut(other_sentence) if self.language == 'chinese' else other_sentence.lower().split())
                  if w not in self.stopwords]

        # 使用Jaccard相似度
        intersection = len(set(words1) & set(words2))
        union = len(set(words1) | set(words2))
        return intersection / union if union != 0 else 0

    def textrank(self, sentences, top_n=5, damping_factor=0.85, max_iter=100):
        """TextRank算法实现"""
        similarity_matrix = np.zeros((len(sentences), len(sentences)))

        # 构建相似度矩阵
        for i in range(len(sentences)):
            for j in range(len(sentences)):
                if i != j:
                    similarity_matrix[i][j] = self.calculate_similarity(sentences[i], sentences[j])

        # 归一化矩阵
        row_sum = similarity_matrix.sum(axis=1)
        normalized_matrix = similarity_matrix / row_sum[:, np.newaxis]

        # 初始化得分
        scores = np.ones(len(sentences))

        # 迭代计算
        for _ in range(max_iter):
            prev_scores = np.copy(scores)
            for i in range(len(sentences)):
                scores[i] = (1 - damping_factor) + damping_factor * np.sum(normalized_matrix[i, :] * prev_scores)

            if np.linalg.norm(scores - prev_scores) < 1e-5:
                break

        # 获取重要句子索引
        ranked_indices = np.argsort(scores)[::-1][:top_n]
        return sorted(ranked_indices)

    def summarize(self, text, ratio=0.2):
        """生成摘要"""
        sentences = self.preprocess_text(text)
        if len(sentences) < 3:
            return "文本过短，无法生成有效摘要"

        top_n = max(1, int(len(sentences) * ratio))
        important_indices = self.textrank(sentences, top_n=top_n)

        # 按原文顺序排列
        selected_sentences = [sentences[i] for i in sorted(important_indices)]

        # 中文使用句号连接，英文使用.连接
        separator = '。' if self.language == 'chinese' else '. '
        return separator.join(selected_sentences) + ('。' if self.language == 'chinese' else '.')


class MainWindow(QMainWindow):
    def __init__(self):
        super().__init__()

        # 初始化摘要器
        self.summarizer = TextRankSummarizer()

        # 界面设置
        self.setup_ui()

    def setup_ui(self):
        self.setWindowTitle("TextRank文本摘要工具")
        self.setGeometry(100, 100, 1000, 800)

        main_widget = QWidget()
        layout = QVBoxLayout()

        # 输入区
        self.input_text = QTextEdit()
        self.input_text.setPlaceholderText("在此粘贴需要摘要的文本（建议500字以上）...")

        # 控制区
        control_layout = QHBoxLayout()
        self.ratio_spin = QSpinBox()
        self.ratio_spin.setRange(5, 50)
        self.ratio_spin.setValue(20)
        self.ratio_spin.setSuffix("%")

        self.lang_combo = QComboBox()
        self.lang_combo.addItems(["中文", "英文"])

        self.summarize_btn = QPushButton("生成摘要")
        self.import_btn = QPushButton("导入文件")
        self.clear_btn = QPushButton("清空")

        control_layout.addWidget(QLabel("摘要比例："))
        control_layout.addWidget(self.ratio_spin)
        control_layout.addWidget(QLabel("语言："))
        control_layout.addWidget(self.lang_combo)
        control_layout.addWidget(self.import_btn)
        control_layout.addWidget(self.summarize_btn)
        control_layout.addWidget(self.clear_btn)

        # 输出区
        self.output_text = QTextEdit()
        self.output_text.setReadOnly(True)

        # 布局组合
        layout.addWidget(QLabel("输入文本："))
        layout.addWidget(self.input_text)
        layout.addLayout(control_layout)
        layout.addWidget(QLabel("摘要结果："))
        layout.addWidget(self.output_text)

        main_widget.setLayout(layout)
        self.setCentralWidget(main_widget)

        # 信号连接
        self.summarize_btn.clicked.connect(self.generate_summary)
        self.import_btn.clicked.connect(self.import_file)
        self.clear_btn.clicked.connect(self.clear_content)
        self.lang_combo.currentTextChanged.connect(self.change_language)

    def change_language(self, lang):
        self.summarizer = TextRankSummarizer('chinese' if lang == "中文" else 'english')

    def generate_summary(self):
        text = self.input_text.toPlainText().strip()
        if not text:
            QMessageBox.warning(self, "输入错误", "请输入需要摘要的文本")
            return

        ratio = self.ratio_spin.value() / 100
        summary = self.summarizer.summarize(text, ratio)
        self.output_text.setPlainText(summary)

    def import_file(self):
        path, _ = QFileDialog.getOpenFileName(
            self, "打开文本文件", "",
            "文本文件 (*.txt);;所有文件 (*.*)"
        )
        if path:
            try:
                with open(path, 'r', encoding='utf-8') as f:
                    self.input_text.setPlainText(f.read())
            except Exception as e:
                QMessageBox.critical(self, "错误", f"文件读取失败：\n{str(e)}")

    def clear_content(self):
        self.input_text.clear()
        self.output_text.clear()


if __name__ == "__main__":
    app = QApplication(sys.argv)
    window = MainWindow()
    window.show()
    sys.exit(app.exec_())