使用NLP库textblob进行情感分析-红楼梦评论

太过爱你忘了你带给我的痛 2022-02-19 23:24 749阅读 0赞

最近做了一个分析国外读者对红楼梦评价的小项目。这部分是使用textblob库对评论进行情感分析，得到情感值，并且进行分类，生成词云。  
生成直方图、条形图的数据分析过程见[我的这篇文章][Link 1]

读入的数据是这样的格式。包含两行，一行评论，一行url来源。  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dqaDI2MjIwNzUxMjc_size_16_color_FFFFFF_t_70]  
生成的结果是这样的result.csv文件  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dqaDI2MjIwNzUxMjc_size_16_color_FFFFFF_t_70 1]  
词云图片：  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dqaDI2MjIwNzUxMjc_size_16_color_FFFFFF_t_70 2]  
代码如下

from textblob import TextBlob
    from wordcloud import WordCloud
    import pandas as pd
    import numpy as np
    import csv
    from os import listdir
    
    def getComments(filename): # 获取评论列表、评论中所有的单词，以空格分隔
        comments = np.zeros(0)
        words = ''
        com_file = pd.read_csv(filename)
        comments = np.append(comments, com_file['comment'])
        for each in comments:
            words += each
        replace_list = [',', '.', '\'', '\"']
        for each in replace_list:
            words = words.replace(each, ' ')
        return comments, words
    
    def getWordCloud(text_str, picture_name): # 生成词云
        wordcloud = WordCloud(background_color="white",width=1980, height=1080, margin=2, random_state=0).generate(text_str)
        wordcloud.to_file(picture_name)
    
    def get_p_or_n(comments): # 获取情绪极化评分，并划定阈值确定是积极、消极或中立
        with open('result.csv', 'w', encoding='utf-8') as csvfile:
            id = 0
            writer = csv.writer(csvfile)
            writer.writerow(['id', 'result', 'score', 'comment'])
            with open('samples.csv', 'w', encoding='utf-8') as samples_file:
                writer_samples = csv.writer(samples_file)
                writer_samples.writerow(['id', 'result', 'score', 'OurJudge', 'comment'])
                for each in comments:
                    judge = TextBlob(each)
                    # print(each)
                    result = ''
                    score = judge.sentiment.polarity
                    if score > 0.05:
                        result = '积极'
                    elif score < -0.03:
                        result = '消极'
                    else:
                        result = '中立'
                    id += 1
                    writer.writerow([id, result, score, each])
                    if id%5 == 0:
                        writer_samples.writerow([id, result, score, '', each])
    
    def main():
        filename = "comments.csv"
        comments, words = getComments(filename)
        print(len(comments))
        getWordCloud(words, "WordCloud.png")
        get_p_or_n(comments)
    
    if __name__ == "__main__":
        main()

[Link 1]: https://blog.csdn.net/wjh2622075127/article/details/89222887
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dqaDI2MjIwNzUxMjc_size_16_color_FFFFFF_t_70]: /images/20220220/1096cba775cc4198a87228420f426bf2.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dqaDI2MjIwNzUxMjc_size_16_color_FFFFFF_t_70 1]: /images/20220220/28cd1b79f4c141b49360b4a0386f263d.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dqaDI2MjIwNzUxMjc_size_16_color_FFFFFF_t_70 2]: /images/20220220/ee8430ce68634eef87a846efdc09a4b2.png