Python弹幕情感分析之《我就是演员》

当我们爬取了弹幕内容后，我们进行简单的情感分析。

import pandas as pd
import jieba
from matplotlib import pyplot as plt
import matplotlib as mpl

mpl.rcParams['font.sans-serif'] = ['SimHei']  # 指定默认字体 SimHei为黑体
mpl.rcParams['axes.unicode_minus'] = False  # 用来正常显示负号

# 加载自定义的词典
jieba.load_userdict('stopwords.txt')
jieba.add_word('韩雪')
jieba.add_word('张钧甯')
jieba.add_word('不用猜')
jieba.add_word('不用想')

pd.set_option('display.max_columns', None)  # pandas在pycharm中显示所有列
data = pd.read_csv('actor_danmu.csv', header=None, names=['id', '点赞数', 'Comment'])  # 设置列名
comment = list(data['Comment'])  # 提取某一列数据，将pandas对象转换为列表

def sent2word(sentence):
    # 分词，并去除停用词
    segList = jieba.cut(sentence, cut_all=False)  # 分词
    segResult = []
    for w in segList:
        segResult.append(w)
    stopwords = open('stopwords.txt', encoding='utf-8-sig').readlines()  # 按行读取，因此每个字符的末尾有换行符
    # 去除每个字符中的换行符
    stop_words = strip_blank(stopwords)
    stop_words.append('不用')  # 添加一个停用词
    newSent = []
    for word in segResult:
        if word in stop_words:
            # print("stopword: %s" % word)  # 可以试着输出我们csv文件中包含的有哪些停用词
            continue
        else:
            newSent.append(word)
    return newSent

def strip_blank(list):
    # 因为是按行读取，所以要去除每个字符中的换行符
    l = []
    for i in list:
        l.append(i.strip('\n'))
    return l

def classify_word(words):
    # 统计各种情感词的个数：正面、负面、否定
    # open()函数打开文件时，设置编码，readlines()方法有换行符，因此要去掉
    positive_list = open('sentiment_word\正面情绪词.txt', encoding='utf-8-sig').readlines()
    negtive_list = open('sentiment_word\负面情绪词.txt', encoding='utf-8-sig').readlines()
    not_list = open('sentiment_word\否定词.txt', encoding='utf-8-sig').readlines()
    positive_list = strip_blank(positive_list)
    negtive_list = strip_blank(negtive_list)
    not_list = strip_blank(not_list)

    number = {}  # 统计三类词汇的数量
    i = j = k =  0
    for num in range(len(words)):
        if words[num] in positive_list:
            i += 1
        elif words[num] in negtive_list:
            j += 1
        elif words[num] in not_list: 
        	k += 1
            if words[num+1] in negtive_list:  # 判断否定词后面的词性
                i += 1
            if words[num + 1] in positive_list:
                j += 1
    number['positive'] = i
    number['negtive'] = j
    number['not'] = k
    return number

def main():
    score = [] # 简单计算情感得分，正面词汇个数减去负面词汇个数
    for i in range(24):  # csv文件中循环24次
        sentence = ''.join(comment[i * 81*2*5: (i + 1) * 81*2*5])
        if sentence:
            words = sent2word(sentence)
            number = classify_word(words)
            score1 = number['positive'] - number['negtive']
            print(score1)
            score.append(score1)
        else:
            break
    y = [i*5 for i in range(1, len(score)+1)]
    plt.plot(y, score)
    
    plt.xlabel('分钟')
    plt.ylabel('情感得分')
    plt.xticks(y)  # 设置坐标轴
    plt.title('我就是演员第十三期之韩雪夺冠弹幕情感分数')
    plt.savefig('我就是演员')  # 图片保存到本地
    plt.show()

if __name__ == '__main__':
    main()

结果如下：