旧代码学习上传记录-b站视频弹幕词频统计画像

首先选择一个弹幕较多，自己喜欢的视频，进行弹幕的采集，比如我当时做项目的时候非常痴迷棋魂，选择了一个棋魂的混剪视频，采集其中的弹幕进行词类分析。

不知道提取相关数据是否是合理操作，所以隐去这段代码，提取的结果如图所示：

 然后根据本数据集，画一个云图的模板，

import numpy as np
import matplotlib.pyplot as plt

x, y = np.ogrid[:300, :300]
mask = (x - 150) ** 2 + (y - 150) ** 2 > 130 ** 2
mask = 255 * mask.astype(int)

plt.imsave('cloud_shape.png', mask, cmap='gray')

然后将这些词的分析填充进模板里面，

import pandas as pd
import jieba
from wordcloud import WordCloud
import matplotlib.pyplot as plt
import numpy as np
from PIL import Image
from snownlp import SnowNLP

excel_file = 'danmaku4.xlsx'
df = pd.read_excel(excel_file)
danmaku_texts = df['弹幕内容'].tolist()


dm_str = " ".join(danmaku_texts)
words_list = jieba.lcut(dm_str)
words_count = {}
for word in words_list:
    words_count[word] = words_count.get(word, 0) + 1

# 输出词频统计
words_freq = sorted(words_count.items(), key=lambda x: x[1], reverse=True)
print("词频统计:")
for word, freq in words_freq:
    print(f"{word}: {freq}")

#掩码
mask_image_path = 'cloud_shape.png'
mask = np.array(Image.open(mask_image_path))  


wordcloud = WordCloud(
    background_color='white',
    mask=mask,                
    width=800,
    height=600,
    max_words=2000,
    max_font_size=100,
    min_font_size=10,
    font_path='C:/Windows/Fonts/simhei.ttf',  
    prefer_horizontal=0.9,      
    relative_scaling=0.5,       
    collocations=False,         # 避免重复单词
    contour_color='steelblue'   
).generate_from_frequencies(words_count)


plt.figure(figsize=(10, 8))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.show()

output_image_path = 'yumu.png'
wordcloud.to_file(output_image_path)
print(f"词云图已保存到 {output_image_path}")
positive = 0
negative = 0
for text in danmaku_texts:
    s = SnowNLP(text)
    if s.sentiments > 0.5:
        positive += 1
    else:
        negative += 1
total = positive + negative
print(f"情感分析结果：")
print(f"积极的弹幕数量: {positive} ({positive / total * 100:.2f}%)")
print(f"消极的弹幕数量: {negative} ({negative / total * 100:.2f}%)")

结果如下图所示：

 输出的结果，借用了snownlp库的函数