102201510/102201510.md

**https://code.educoder.net/p4fmntoqa/102201510**

# 一、PSP表格
| **PSP2.1**                              | **Personal Software Process Stages**    | **预估耗时（分钟）**  | **实际耗时（分钟）** |
| :-------------------------------------- | --------------------------------------- | -------------------- | -------------------- |
| Planning                                | 计划                                    |     20               |          20           |
| · Estimate                              | · 估计这个任务需要多少时间                |     1200             |         1440         |
| Development                             | 开发                                    |     1320              |         1440         |
| · Analysis                              | · 需求分析 (包括学习新技术)               |     360              |         300          |
| · Design Spec                           | · 生成设计文档                           |     120              |         90           |
| · Design Review                         | · 设计复审                               |     120              |         180          |
| · Coding Standard                       | · 代码规范 (为目前的开发制定合适的规范)    |     60               |         60           |
| · Design                                | · 具体设计                              |      180              |         240          |
| · Coding                                | · 具体编码                              |      240              |         270          |
| · Code Review                           | · 代码复审                              |      120              |         180          |
| · Test                                  | · 测试（自我测试，修改代码，提交修改）     |      120              |         120          |
| Reporting                               | 报告                                    |      240              |         210         |
| · Test Repor                            | · 测试报告                              |      60               |          60         |
| · Size Measurement                      | · 计算工作量                            |      60               |          60         |
| · Postmortem & Process Improvement Plan | · 事后总结, 并提出过程改进计划            |      120              |          90         |
|                                         | · 合计                                  |      1580             |         1670         |
# 二、任务要求的实现
## 1. 项目设计与技术栈
#### 从阅读完题目到完成作业，这一次的任务被你拆分成了几个环节？你分别通过什么渠道、使用什么方式方法完成了各个环节？列出你完成本次任务所使用的技术栈。（5'）
**我个人将本次任务拆为了四个部分，分别是弹幕数据获取、数据处理与转存、词云图绘制以及附加的弹幕情绪分析。**
1) 弹幕数据获取主要是通过在B站上学习爬虫内容，查阅各种爬虫文件，辅以浏览器自带的抓包工具，最终编写出了自己的b站爬虫程序。
2) 数据处理和转存大部分内容都是通过我个人之前的知识积累，自行完成的，部分难题则通过查找导入的第三方库函数说明文档解决。 
3) 词云图绘制仅有绘制词云图的部分去网络上搜索了相关的内容，剩余部分则是以自己的能力编写的内容。
4) 弹幕情绪分析是通过网络上下载相关的自然语言分析模型，进行对应的情感分析，并最终输出成柱状图的形式，模型的调用部分参考了模型的说明文档。

**用到的技术栈有如下内容：**
1) <font size=3 color='gray' face="黑体">文本处理和自然语言处理</font>:
    *jieba: 中文分词库
    paddlenlp: 自然语言处理库*
2) <font size=3 color='gray' face="黑体">数据分析和科学计算</font>:
    *numpy: 数值计算库
    pandas: 数据处理和分析库
    scikit_learn: 机器学习库*
3) <font size=3 color='gray' face="黑体">数据可视化</font>:
    *matplotlib: 数据可视化库
    wordcloud: 词云生成库*
4) <font size=3 color='gray' face="黑体">文件处理</font>:
    *openpyxl: Excel文件处理库
    Pillow: 图像处理库*
5) <font size=3 color='gray' face="黑体">深度学习</font>:
    *paddlepaddle: 深度学习框架*
6) <font size=3 color='gray' face="黑体">网络请求</font>:
    *Requests: HTTP请求库*
7) <font size=3 color='gray' face="黑体">开发工具</font>:
    *pipreqs: 自动生成requirements文件
    unittest: 单元测试框架
    coverage: 代码覆盖率测试工具
    viztracer: 代码性能分析工具
    git: 版本控制工具*

## 2. 爬虫与数据处理
#### 说明业务逻辑，简述代码的设计过程（例如可介绍有几个类，几个函数，他们之间的关系），并对关键的函数或算法进行说明。（20'）
本程序主要功能是对b站相关视频的弹幕进行数据处理和分析，全程共用到了15个库及其库函数，自定义了13个函数以完成程序的功能设计。程序的四个部分均有一个较为主要的函数。以下是对主要函数的说明：
1) **弹幕数据获取中的正则匹配函数re.findall()。** 该部分的所有函数几乎都用到了re.findall()，所有有效信息的获取也都离不开re.findall()。它一共需要两个参数，一个是匹配字段文本，一个是数据文本，返回值是list形式的匹配字段。
2) **数据处理与转存中的数据类型转换函数ChangeDfToString()。** 该部分的后续函数都是基于ChangeDfToString()转换出来的字符串进行进行处理的。用dataframe正常转换出来字符串会有很多的空格以及莫名其妙的字符，通过这个函数可以去除无效的空格，并对每一个有效内容加以逗号分隔。它一共可以接受4个参数，原dataframe，分隔字符sep，是否保存标志isSave，保存路径filePath，返回值是分隔好的字符。
3) **词云绘制部分的词频转换函数ChangeToFreq()。** 该函数可以将分隔好的词语用TF-IDF关键词提取，并统计各类词语的词频，实现正则化，对后续的词云图绘制有较大的帮助。它接受一个参数，分隔好的字符串，返回值是词组频率的字典，键值对是词组和频率。
4) **弹幕情绪分析部分的模型加载类Taskflow()。** 该类是paddlenlp库中的Taskflow类。panddnlp是基于百度的飞浆平台搭建的自然语言处理（NLP）模型库，对中文语言分析有非常优秀的表现。利用Taskflow()类，可以搭载模型model，设定语言处理的模式schema，最后会返回一个ie模型对象，之后就可以使用ie对文本进行语言处理了。

## 3. 数据统计接口部分的性能改进
#### 记录在数据统计接口的性能上所花费的时间，描述你改进的思路，并展示一张性能分析图（例如可通过VS /JProfiler的性能分析工具自动生成），并展示你程序中消耗最大的函数。（6'）
由于B站反爬虫的机制，为了防止被ban，我对程序自行设定了一定的延时，该部分就是所需时间最长的一部分函数。
其他的函数性能消耗都不会很大，除了使用ie模型对文本进行语言分析处理。**针对此类性能消耗，我缩小了文本量，从全文本处理改成了随机抽样处理，数据结果相差不大，但性能消耗下降了非常多。**
以下是除了b站爬虫程序以外的其他程序性能消耗：
![avatar][img_1]
<font size=2><center>*可以看到性能消耗最大的函数是ChangeDfToString()，其余函数的性能消耗都被挤到了角落*</center></font>
![avatar][img_2]
<font size=2><center>*可以看到显示性能消耗最大的函数是CreateWordCloud()，但这个函数会绘制图片，关闭图片才结束统计，有一定的误差*</center></font>
![avatar][img_3]
<font size=2><center>*可以看到显示性能消耗最大的是loadModel()函数后面的一串点点点，这里的每一个部分都是一次文本模型处理，性能开销非常大*</center></font>
![avatar][img_6]
<font size=2><center>*这里展示的是代码的单元测试及覆盖率的内容，因为找不到适合的地方就放在这里了*</center></font>

## 4. 数据结论的可靠性
#### 介绍结论的内容，以及通过什么数据以及何种判断方式得出此结论（6'）
本程序得出的结论比较多，以下是结论介绍和相关的数据判断：
1) **b站用户对于弹幕讨论巴黎奥运会的ai使用并不感兴趣。** 统计了b站搜索“2024巴黎奥运会”综合排序前300的视频弹幕，并对弹幕应用了ai、智能等关键字提取，最终发现基本没几个相关弹幕，弹幕数量最多的还是“该内容疑似使用AI技术合成，请谨慎甄别”，不能说和巴黎奥运会没有关系，只能说和巴黎奥运会毫不相关。
2) **b站用户在弹幕讨论中，正向情绪远大于负向情绪。** 不仅可以从词云图中看出，b站用户在巴黎奥运会相关视频发送的最多的弹幕是“哈哈哈”、“好看”、“加油”等等积极正向的词语；也能通过后续的弹幕情绪分析中看出，1000个弹幕中，正向情绪的弹幕有658个，负向情绪的弹幕有341个，正向接近负向的两倍，而且，弹幕的情绪可信度也非常高，正向与负向的情绪可信度均接近90%。
![avatar][img_4]
<font size=2><center>*弹幕词云图*</center></font>
![avatar][img_5]
<font size=2><center>*弹幕情绪倾向图*</center></font>

## 5. 数据可视化界面的展示
#### 在博客中介绍数据可视化界面的组件和设计的思路。（15'）
数据可视化主要是通过matplotlib: 数据可视化库、wordcloud: 词云生成库以及Pillow: 图像处理库函数来实现的。
1) 考虑到词云生成为**默认的长方形不够美观**，因此**特地从网络上找到了巴黎奥运会吉祥物的标志性图片，并将词云图中的文字按图片颜色排版**，最终得到了现在的词云图片，观感好上了不少。
2) 虽然我利用了自然文本处理模型，对弹幕文本进行了情绪倾向分析，但这些都只是文本的内容。个人认为，单纯的文本内容表述不足以给出直观的数据展示，因此我将这部分内容以**信息量最大、比较直观的正向负向柱形图**的形式展现出来。

# 三、心得体会
#### 在这儿写下你完成本次作业的心得体会，当然，如果你还有想表达的东西但在上面两个板块没有体现，也可以写在这儿~（10'）
本次作业是我**第一次尝试以python完成一个这么大类的一个程序**，也是我**第一次接触到项目管理中的单元测试、代码覆盖率等待程序性能优化相关的内容**，受益匪浅。不论是代码编写能力还是程序的设计能力，都有了足量的提升。
不过这里想吐槽一下 <font size=3>**~~在2024巴黎奥运会的视频中寻找ai，真的没有搞错什么吗？~~**</font>（爬虫爬得都不自信了）























[img_1]:data:image/png;base64,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
[img_2]:data:image/png;base64,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
[img_3]:data:image/png;base64,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
[img_4]:data:image/png;base64,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
[img_5]:data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAE4gAAA2sCAYAAAD2RCd4AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjkuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8hTgPZAAAACXBIWXMAAEzlAABM5QF1zvCVAAEAAElEQVR4nOzde5TVZb348c/ch5kBVEpMUEFFNM1IKTEtSbxrombeSi2sLBcWlh4FtDI1bJkWKRmmqZkieDlqrlQkA0XFErTTSUUEUsBUCBHnwtx/f3iYnzh7w8wws+cZfL3W2ms1z7O/z/7MsGfvWau93uY1Nzc3BwAAAAAAAAAAAAAAAAAAAAAAAADdLr+7BwAAAAAAAAAAAAAAAAAAAAAAAADgPQJxAAAAAAAAAAAAAAAAAAAAAAAAAIkQiAMAAAAAAAAAAAAAAAAAAAAAAABIhEAcAAAAAAAAAAAAAAAAAAAAAAAAQCIE4gAAAAAAAAAAAAAAAAAAAAAAAAASIRAHAAAAAAAAAAAAAAAAAAAAAAAAkAiBOAAAAAAAAAAAAAAAAAAAAAAAAIBECMQBAAAAAAAAAAAAAAAAAAAAAAAAJEIgDgAAAAAAAAAAAAAAAAAAAAAAACARAnEAAAAAAAAAAAAAAAAAAAAAAAAAiRCIAwAAAAAAAAAAAAAAAAAAAAAAAEiEQBwAAAAAAAAAAAAAAAAAAAAAAABAIgTiAAAAAAAAAAAAAAAAAAAAAAAAABIhEAcAAAAAAAAAAAAAAAAAAAAAAACQCIE4AAAAAAAAAAAAAAAAAAAAAAAAgEQIxAEAAAAAAAAAAAAAAAAAAAAAAAAkQiAOAAAAAAAAAAAAAAAAAAAAAAAAIBECcQAAAAAAAAAAAAAAAAAAAAAAAACJEIgDAAAAAAAAAAAAAAAAAAAAAAAASIRAHAAAAAAAAAAAAAAAAAAAAAAAAEAiBOIAAAAAAAAAAAAAAAAAAAAAAAAAEiEQBwAAAAAAAAAAAAAAAAAAAAAAAJAIgTgAAAAAAAAAAAAAAAAAAAAAAACARAjEAQAAAAAAAAAAAAAAAAAAAAAAACRCIA4AAAAAAAAAAAAAAAAAAAAAAAAgEQJxAAAAAAAAAAAAAAAAAAAAAAAAAIkQiAMAAAAAAAAAAAAAAAAAAAAAAABIhEAcAAAAAAAAAAAAAAAAAAAAAAAAQCIE4gAAAAAAAAAAAAAAAAAAAAAAAAASIRAHAAAAAAAAAAAAAAAAAAAAAAAAkAiBOAAAAAAAAAAAAAAAAAAAAAAAAIBECMQBAAAAAAAAAAAAAAAAAAAAAAAAJEIgDgAAAAAAAAAAAAAAAAAAAAAAACARAnEAAAAAAAAAAAAAAAAAAAAAAAAAiRCIAwAAAAAAAAAAAAAAAAAAAAAAAEiEQBwAAAAAAAAAAAAAAAAAAAAAAABAIgTiAAAAAAAAAAAAAAAAAAAAAAAAABIhEAcAAAAAAAAAAAAAAAAAAAAAAACQCIE4AAAAAAAAAAAAAAAAAAAAAAAAgEQIxAEAAAAAAAAAAAAAAAAAAAAAAAAkQiAOAAAAAAAAAAAAAAAAAAAAAAAAIBECcQAAAAAAAAAAAAAAAAAAAAAAAACJEIgDAAAAAAAAAAAAAAAAAAAAAAAASIRAHAAAAAAAAAAAAAAAAAAAAAAAAEAiBOIAAAAAAAAAAAAAAAAAAAAAAAAAEiEQBwAAAAAAAAAAAAAAAAAAAAAAAJAIgTgAAAAAAAAAAAAAAAAAAAAAAACARAjEAQAAAAAAAAAAAAAAAAAAAAAAACRCIA4AAAAAAAAAAAAAAAAAAAAAAAAgEQJxAAAAAAAAAAAAAAAAAAAAAAAAAIkQiAMAAAAAAAAAAAAAAAAAAAAAAABIhEAcAAAAAAAAAAAAAAAAAAAAAAAAQCIE4gAAAAAAAAAAAAAAAAAAAAAAAAASIRAHAAAAAAAAAAAAAAAAAAAAAAAAkAiBOAAAAAAAAAAAAAAAAAAAAAAAAIBECMQBAAAAAAAAAAAAAAAAAAAAAAAAJEIgDgAAAAAAAAAAAAAAAAAAAAAAACARAnEAAAAAAAAAAAAAAAAAAAAAAAAAiRCIAwAAAAAAAAAAAAAAAAAAAAAAAEiEQBwAAAAAAAAAAAAAAAAAAAAAAABAIgTiAAAAAAAAAAAAAAAAAAAAAAAAABIhEAcAAAAAAAAAAAAAAAAAAAAAAACQCIE4AAAAAAAAAAAAAAAAAAAAAAAAgEQIxAEAAAAAAAAAAAAAAAAAAAAAAAAkQiAOAAAAAAAAAAAAAAAAAAAAAAAAIBECcQAAAAAAAAAAAAAAAAAAAAAAAACJEIgDAAAAAAAAAAAAAAAAAAAAAAAASIRAHAAAAAAAAAAAAAAAAAAAAAAAAEAiBOIAAAAAAAAAAAAAAAAAAAAAAAAAEiEQBwAAAAAAAAAAAAAAAAAAAAAAAJAIgTgAAAAAAAAAAAAAAAAAAAAAAACARAjEAQAAAAAAAAAAAAAAAAAAAAAAACRCIA4AAAAAAAAAAAAAAAAAAAAAAAAgEQJxAAAAAAAAAAAAAAAAAAAAAAAAAIkQiAMAAAAAAAAAAAAAAAAAAAAAAABIhEAcAAAAAAAAAAAAAAAAAAAAAAAAQCIE4gAAAAAAAAAAAAAAAAAAAAAAAAASIRAHAAAAAAAAAAAAAAAAAAAAAAAAkAiBOAAAAAAAAAAAAAAAAAAAAAAAAIBECMQBAAAAAAAAAAAAAAAAAAAAAAAAJEIgDgAAAAAAAAAAAAAAAAAAAAAAACARAnEAAAAAAAAAAAAAAAAAAAAAAAAAiRCIAwAAAAAAAAAAAAAAAAAAAAAAAEiEQBwAAAAAAAAAAAAAAAAAAAAAAABAIgTiAAAAAAAAAAAAAAAAAAAAAAAAABIhEAcAAAAAAAAAAAAAAAAAAAAAAACQCIE4AAAAAAAAAAAAAAAAAAAAAAAAgEQIxAEAAAAAAAAAAAAAAAAAAAAAAAAkQiAOAAAAAAAAAAAAAAAAAAAAAAAAIBECcQAAAAAAAAAAAAAAAAAAAAAAAACJEIgDAAAAAAAAAAAAAAAAAAAAAAAASIRAHAAAAAAAAAAAAAAAAAAAAAAAAEAiBOIAAAAAAAAAAAAAAAAAAAAAAAAAEiEQBwAAAAAAAAAAAAAAAAAAAAAAAJAIgTgAAAAAAAAAAAAAAAAAAAAAAACARAjEAQAAAAAAAAAAAAAAAAAAAAAAACRCIA4AAAAAAAAAAAAAAAAAAAAAAAAgEQJxAAAAAAAAAAAAAAAAAAAAAAAAAIkQiAMAAAAAAAAAAAAAAAAAAAAAAABIhEAcAAAAAAAAAAAAAAAAAAAAAAAAQCIE4gAAAAAAAAAAAAAAAAAAAAAAAAASIRAHAAAAAAAAAAAAAAAAAAAAAAAAkAiBOAAAAAAAAAAAAAAAAAAAAAAAAIBECMQBAAAAAAAAAAAAAAAAAAAAAAAAJEIgDgAAAAAAAAAAAAAAAAAAAAAAACARAnEAAAAAAAAAAAAAAAAAAAAAAAAAiRCIAwAAAAAAAAAAAAAAAAAAAAAAAEiEQBwAAAAAAAAAAAAAAAAAAAAAAABAIgTiAAAAAAAAAAAAAAAAAAAAAAAAABIhEAcAAAAAAAAAAAAAAAAAAAAAAACQCIE4AAAAAAAAAAAAAAAAAAAAAAAAgEQIxAEAAAAAAAAAAAAAAAAAAAAAAAAkQiAOAAAAAAAAAAAAAAAAAAAAAAAAIBECcQAAAAAAAAAAAAAAAAAAAAAAAACJEIgDAAAAAAAAAAAAAAAAAAAAAAAASIRAHAAAAAAAAAAAAAAAAAAAAAAAAEAiBOIAAAAAAAAAAAAAAAAAAAAAAAAAEiEQBwAAAAAAAAAAAAAAAAAAAAAAAJAIgTgAAAAAAAAAAAAAAAAAAAAAAACARAjEAQAAAAAAAAAAAAAAAAAAAAAAACRCIA4AAAAAAAAAAAAAAAAAAAAAAAAgEQJxAAAAAAAAAAAAAAAAAAAAAAAAAIkQiAMAAAAAAAAAAAAAAAAAAAAAAABIhEAcAAAAAAAAAAAAAAAAAAAAAAAAQCIE4gAAA
[img_6]:data:image/png;base64,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