Python Data Preprocessing and Visualization of Jay Chou Lyrics: From JSON to Word Cloud

The example uses a Chinese lyric database stored in JSON format. First, the JSON file can be converted to a CSV or XLSX file with an online converter, then filtered in Excel to keep only songs by Jay Chou and export the lyrics to plain text.

Alternatively, the entire preprocessing can be scripted in Python. The required libraries are imported with:

import json

The JSON file is loaded:

with open('lyrics.json', 'r') as f:
    data = json.load(f)

Jay Chou entries are extracted:

data_zjl = [item for item in data if item['singer'] == '周杰伦']
print(len(data_zjl))

All lyrics are collected into a list and written to a text file:

Zjl_lyrics = []
for song in data_zjl:
    Zjl_lyrics = Zjl_lyrics + song['lyric']

with open('zjl_lyrics.txt', 'w') as outfifile:
    outfifile.write('\n'.join(Zjl_lyrics))

For word segmentation, Jieba, pandas, and Counter are used. A stop‑word list is loaded first:

with open('chinese_stop_words.txt') as f:
    stopwords = [line.strip() for line in f.readlines()]

The lyric file is read, segmented, cleaned, and frequency counted:

fifile = open('zjl_lyrics.txt').read()
words = jieba.lcut(fifile, cut_all=False, use_paddle=True)
words = [w for w in words if w not in stopwords]
words = [w.strip() for w in words]
words = [w for w in words if w != ' ']
words_fifilter = [w for w in words if len(w) > 1]
df = pd.DataFrame.from_dict(Counter(words_fifilter), orient='index').reset_index()
df = df.rename(columns={'index': 'words', 0: 'count'})
df.to_excel('周杰伦分词结果.xlsx')

The resulting Excel file contains each word and its frequency, which can be visualized. A word cloud is generated with the WordCloud library:

from wordcloud import WordCloud
wc = WordCloud(font_path='Alibaba-PuHuiTi-Regular.ttf', background_color='white', max_words=2000)
wc.generate(' '.join(words_fifilter))
import matplotlib.pyplot as plt
plt.imshow(wc)
plt.figure(figsize=(12,10), dpi=300)
plt.axis('off')
plt.show()

Alternatively, online tools such as 微词云, 易词云, or 图悦 can import the Excel file (columns named "单词" and "词频") to produce customizable word clouds, allowing users to filter by part of speech, adjust font size, colors, and mask shapes (e.g., a portrait of Jay Chou).

Other visualizations like proportional circles can be created by importing the frequency data into chart tools and generating pie‑style diagrams.

The guide emphasizes that different segmentation methods may yield varying results, so readers should compare approaches to choose the most suitable one for their analysis.