博文

AGI-AIGC-GPT测评DIKWP（全球）实验室（测评系列之十）-世界上第一篇大模型安全领域研究报告

已有 1951 次阅读 2023-4-19 17:48 |系统分类:论文交流

The Capability Evaluation of ChatGPT, GPT-4 ,Tongyi Qianwen on Security Domain with DIKWP Analysis AGI-AIGC-GPT Test and Evaluation DIKWP (Global) Lab

April 2023

DOI: 10.13140/RG.2.2.24611.12324

https://www.researchgate.net/publication/370106437_The_Capability_Evaluation_of_ChatGPT_GPT-4_Tongyi_Qianwen_on_Security_Domain_with_DIKWP_Analysis_AGI-AIGC-GPT_Test_and_Evaluation_DIKWP_Global_Lab

AGI-AIGC-GPT测评DIKWP（全球）实验室（测评报告系列之十）

The Capability Evaluation of ChatGPT, GPT-4 ,Tongyi Qianwen on Security Domain with DIKWP Analysis

ChatGPT, GPT-4 , 通义千问的安全领域DIKWP能力评测报告

Yanfei Liu¹^{, 2}, Kunguang Wu⁴^,5_,Yucong Duan*⁴^,5, Wentao Wang³

cqlyf@tju.edu.cn¹^,², duanyucong@hotmail.com⁴^,5

College of Intelligence and Computing, Tianjin University¹

Chongqing Police Academy²

Chongqing University of Technology³

AGI-AIGC-GPT Test and Evaluation DIKWP (Global) Lab⁴

DIKWP research group, Hainan University⁵

摘要： 本文基于ChatGPT、GPT-4和通义千问分别开展了网络空间安全领域个案DIKWP分析，集合案DIKWP分析以及专家角色下的集合案DIKWP分析，并通过GPT-4进行了两类DIKWP差异图谱比较，最后得出结论：网络安全领域，DIKWP模式在先验知识的加持下有更好的转化与互补处理能力和校验能力。

关键词：ChatGPT，GPT4.0，网络安全， DIKWP评测

1. 前言

自然语言处理技术在网络安全领域的应用越来越成为关注的话题。ChatGPT\GPT-4以其强大的文本生成和理解能力吸引了越来越多研究者的关注。本文旨在以六个典型网络安全案例为基础，探讨ChatGPT\GPT-4在网络安全领域的应用效果，并利用DIKWP评测指标进行评估，从而深入分析ChatGPT\GPT-4应用的优势与劣势。文章通过分析各种图谱类型在该领域中的作用，旨在为读者提供一个全面的认识，以期推动该领域中的人工智能应用研究。

2. 评测

本团队基于DIKWP框架^[¹^],利用LLM展示了利用DIKWP来处理复杂问题^[²^-3^]，以下是关于网络安全方向上的DIKWP测评。我们使用了一些基于网络安全案件的新闻针对ChatGPT与GPT-4进行DIKWP评测，通过构建出数据图谱、信息图谱、知识图谱、智慧图普以及意图图谱来分析其处理能力。

2.1 基于单个案例的内容评测

在本节测试中，我们将6个关于网络空间安全相关的案件作为测试用例分别输入到ChatGPT和GPT-4中，使其针对案件的内容输出对应的数据图谱、信息图谱、知识图谱、智慧图谱和意图图谱。并比较其两者在DIKWP测试上对网络空间安全相关的案件的分析能力。

2.1.1案例1

1) 案例标题：多个团伙利用社保公积金系统漏洞，非法获取公民个人信息2300万条。

2) 案例内容：2022年12月7日，四川南充市公安局顺庆区分局侦破一起公安部挂牌督办案件，抓获犯罪嫌疑人121人，查获公民个人信息2300余万条，发现国内多地各类信息系统平台漏洞300余个，收缴黑客工具12套。

3) 输入内容：“2022年12月7日，四川南充市公安局顺庆区分局侦破一起公安部挂牌督办案件，抓获犯罪嫌疑人121人，查获公民个人信息2300余万条，发现国内多地各类信息系统平台漏洞300余个，收缴黑客工具12套。” 请根据上述内容，请分别提取出数据图谱、信息图谱、知识图谱、智慧图谱、意图图谱。

4) 回答内容：

a) ChatGPT回答内容：

以下是可能提取的各种图谱：

数据图谱：这段内容中包含很多数字数据，例如抓获嫌疑人数量、信息条目数量、漏洞数量等，这些数字可以构成数据图谱，帮助更好地理解案件背景和情况。

信息图谱：案件中涉及到很多关键词，如“公安部挂牌督办案件”、“犯罪嫌疑人”、“公民个人信息”、“信息系统平台漏洞”等，这些关键词可以构成信息图谱，呈现关键词之间的关系和层次结构。

知识图谱：由于案件涉及到很多技术、安全、法律等方面的知识，在处理这些信息时可以通过构建知识图谱来呈现这些知识之间的关系和层次结构，帮助更好地理解案件相关的知识和概念。

智慧图谱：这个案件中，公安局的侦破行动和技术处理过程中，涉及到很多智慧和经验，如调查技巧、安全技术、应急处置等方面，这些经验和智慧可以构成智慧图谱，帮助更好地记录和共享这些经验和智慧。

意图图谱：这个案件中，在嫌疑人的个人信息被窃取之后，可能涉及到很多网络欺诈、身份盗窃等方面的犯罪行为，为了更好地预防这些行为，可以构建意图图谱，帮助预测这些行为的可能性和模式。

b) GPT-4回答内容：

根据上述内容，我们可以将信息提取为以下几类图谱：

数据图谱：

日期：2022年12月7日

地点：四川南充市公安局顺庆区分局

犯罪嫌疑人数量：121人

查获公民个人信息：2300余万条

发现信息系统平台漏洞：300余个

收缴黑客工具：12套

信息图谱：

事件：侦破一起公安部挂牌督办案件

涉及的行业：网络安全、个人信息保护

案件类型：非法获取、出售公民个人信息

知识图谱：

涉及的法律法规：《中华人民共和国网络安全法》、《中华人民共和国刑法》等

技术手段：黑客工具、漏洞利用

网络犯罪特征：涉及大量公民个人信息、跨地区犯罪

智慧图谱：

防范策略：加强信息系统安全防护、提高公民个人信息保护意识

执法部门：公安部、地方公安局

执法手段：侦查、抓捕、取证、起诉

意图图谱：

犯罪嫌疑人意图：非法牟利，通过获取并出售公民个人信息

执法部门意图：捣毁犯罪团伙，保护公民个人信息安全，维护网络空间安全

社会意图：提高对网络犯罪的警惕，保护自身信息安全

c) 通义千问回答内容：

作为一个人工智能语言模型，目前我还没有学会如何回答这个问题，我会持续学习，为您提供更好的服务

2.1.2案例2

1) 案例标题：夫妻俩贩卖寄递用户个人信息获利24万余元。

2) 案例内容：男子利用从事寄递客服的便利条件，伙同妻子查询、出售大量寄递用户的个人信息牟利，非法获利24万余元，两人双双获刑。2022年11月23日，最高人民检察院发布检察机关落实“七号检察建议”典型案例，其中包括这起“江苏王某涛、董某婷侵犯公民个人信息案”。

3) 输入内容：男子利用从事寄递客服的便利条件，伙同妻子查询、出售大量寄递用户的个人信息牟利，非法获利24万余元，两人双双获刑。2022年11月23日，最高人民检察院发布检察机关落实“七号检察建议”典型案例，其中包括这起“江苏王某涛、董某婷侵犯公民个人信息案”。男子利用从事寄递客服的便利条件，伙同妻子查询、出售大量寄递用户的个人信息牟利，非法获利24万余元，两人双双获刑。2022年11月23日，最高人民检察院发布检察机关落实“七号检察建议”典型案例，其中包括这起“江苏王某涛、董某婷侵犯公民个人信息案”。请根据上述内容，请分别提取出数据图谱、信息图谱、知识图谱、智慧图谱、意图图谱。

4) 回答内容：

a) ChatGPT回答内容：

数据图谱：这段内容中包含一些数字数据，如非法获利的金额24万余元等，这些数字可以构成数据图谱，帮助更好地了解案件的经济影响和涉及的量。

信息图谱：该案中涉及到一对夫妻的名字，地点、犯罪行为、最高人民检察院发布的检察机关落实“七号检察建议”等关键词，这些关键词可以构成信息图谱，帮助更好地了解案件的主要内容和相关方面。

知识图谱：在具体案件中，涉及到很多有关个人信息保护的法律、法规、政策和操作流程等方面的知识，这些可能被组织成知识图谱，以便更好地理解和学习与个人信息保护相关的知识。

智慧图谱：在案件调查和审判过程中，涉及到一定的技术、调查技巧、审判技巧、表达技巧等智慧和实践经验。可以通过构建智慧图谱，可帮助更好地记录和分享这些智慧和经验。

意图图谱：在个人信息保护方面，可能会涉及到很多欺诈、盗窃和其他形式的违法犯罪行为。可以通过构建意图图谱，对这些行为可能性和模式进行预测和分析，帮助更好地预防和打击此类犯罪。

b) GPT-4回答内容：

根据上述内容，我们可以提取以下关键信息：

数据图谱：

日期：2022年11月23日

地点：江苏

涉案人：王某涛、董某婷

非法获利：24万余元

信息图谱：

事件：侵犯公民个人信息案

涉及行业：寄递服务

案件类型：非法查询、出售寄递用户个人信息

知识图谱：

涉及的法律法规：《中华人民共和国网络安全法》、《中华人民共和国刑法》等

犯罪手段：利用寄递客服职务便利，查询、出售用户个人信息

智慧图谱：

防范策略：加强寄递行业对个人信息的保护，提高公民个人信息保护意识

执法部门：最高人民检察院

执法手段：侦查、起诉、判决

意图图谱：

犯罪嫌疑人意图：非法牟利，通过查询、出售寄递用户的个人信息

执法部门意图：惩处侵犯公民个人信息犯罪，保护公民个人信息安全

社会意图：提高对侵犯个人信息犯罪的警惕，加强个人信息保护

c) 通义千问回答内容：

数据图谱：该案例涉及的数据包括姓名、电话号码、住址等公民个人信息。将这些信息以图形化方式展示可以得到一个数据图谱。数据图谱以节点表示个人信息，边表示个人信息之间的联系，节点和边的颜色、形状等可以反映出不同的属性值。这个数据图谱可以帮助人们更直观地了解该案例中个人信息的分布情况，以及它们之间的关系。

信息图谱：该案例涉及的信息包括姓名、电话号码、住址等公民个人信息，以及非法获利的金额等。将这些信息以图形化方式展示可以得到一个信息图谱。信息图谱以节点表示个人信息和非法获利的数据，边表示它们之间的关系。这个信息图谱可以帮助人们更清晰地了解该案例中各种信息的来源和关系，以及非法获利的规模和范围。

知识图谱：该案例涉及的知识包括公民个人信息保护、网络犯罪等。将这些知识以图形化方式展示可以得到一个知识图谱。知识图谱以节点表示知识和关系，边表示知识和知识之间的联系。这个知识图谱可以帮助人们更全面地了解该案例中的相关知识和概念，以及它们之间的关系。

智慧图谱：该案例涉及的智慧包括数据安全、网络安全等。将这些智慧以图形化方式展示可以得到一个智慧图谱。智慧图谱以节点表示智慧和关系，边表示智慧和智慧之间的联系。这个智慧图谱可以帮助人们更深入地了解该案例中的相关智慧和技术，以及它们之间的关系。

意图图谱：该案例涉及的意图包括非法获利和侵犯公民个人信息。将这些意图以图形化方式展示可以得到一个意图图谱。意图图谱以节点表示意图和关系，边表示意图和意图之间的联系。这个意图图谱可以帮助人们更准确地了解该案例中各个意图的目的和意义，以及它们之间的关系。

2.1.3案例3

1) 案例标题：涉及27个省份的特大系列侵犯公民个人信息案件。

2) 案例内容：2022年9月22日，鸡西市公安局侦破特大侵犯公民个人信息案件，共抓获犯罪嫌疑人322名，依法扣押涉案资金560余万元，扣押电脑78部，手机412部，涉案银行卡638张。

3) 输入内容：“2022年9月22日，鸡西市公安局侦破特大侵犯公民个人信息案件，共抓获犯罪嫌疑人322名，依法扣押涉案资金560余万元，扣押电脑78部，手机412部，涉案银行卡638张。”2022年9月22日，鸡西市公安局侦破特大侵犯公民个人信息案件，共抓获犯罪嫌疑人322名，依法扣押涉案资金560余万元，扣押电脑78部，手机412部，涉案银行卡638张。请根据上述内容，请分别提取出数据图谱、信息图谱、知识图谱、智慧图谱、意图图谱。

4) 回答内容：

a) ChatGPT回答内容：

数据图谱：涉及到的数字数据很多，如抓获的犯罪嫌疑人数量、涉案资金金额、电脑、手机和银行卡的数量等，这些数字可以构成数据图谱，便于更好地了解案件的规模、影响范围和损失程度。

信息图谱：案件涉及到很多地区和部门，如27个省份、鸡西市公安局等，以及涉案资金、电脑、手机和银行卡等关键词，这些关键词可以构成信息图谱，呈现其关系和层次结构。

知识图谱：在涉及大规模个人信息泄露和利用的案件中，涉及到很多个人信息保护和安全、网络安全、法律法规等方面的知识，这些知识可以被组织成知识图谱，以便更好地理解和学习与个人信息保护相关的知识。

智慧图谱：在推进此类案件中，涉及到很多技术和操作经验，如调查技巧、取证方法、数据挖掘技术等，这些经验和智慧可以构建智慧图谱，帮助更好地记录和分享这些经验和智慧。

意图图谱：在个人信息保护方面，可能会涉及到很多网络欺诈、身份盗窃等违法犯罪行为。可以通过构建意图图谱，在宏观层面上预测和分析这些行为的可能性和模式，从而帮助更好地预防和打击此类犯罪。

b) GPT-4回答内容：

根据上述内容，我们可以提取以下关键信息：

数据图谱：

日期：2022年9月22日

地点：鸡西市公安局

犯罪嫌疑人数量：322名

扣押涉案资金：560余万元

扣押电脑：78部

扣押手机：412部

涉案银行卡：638张

信息图谱：

事件：侦破特大侵犯公民个人信息案件

涉及行业：网络安全、个人信息保护

案件类型：非法获取、出售公民个人信息

知识图谱：

涉及的法律法规：《中华人民共和国网络安全法》、《中华人民共和国刑法》等

技术手段：电脑、手机等设备

网络犯罪特征：涉及大量公民个人信息、涉及大量资金

智慧图谱：

防范策略：加强信息系统安全防护、提高公民个人信息保护意识

执法部门：鸡西市公安局

执法手段：侦查、抓捕、扣押涉案资金和设备

意图图谱：

犯罪嫌疑人意图：非法牟利，通过获取并出售公民个人信息

执法部门意图：捣毁犯罪团伙，保护公民个人信息安全，维护网络空间安全

社会意图：提高对网络犯罪的警惕，保护自身信息安全

c) 通义千问回答内容：

作为一个人工智能语言模型，目前我还没有学会如何回答这个问题，我会持续学习，为您提供更好的服务。

2.1.4案例4

1) 案例标题：国家互联网信息办公室对滴滴公司依法作出网络安全审查相关行政处罚的决定

2) 案例内容：根据网络安全审查结论及发现的问题和线索，国家互联网信息办公室（以下简称“国家网信办”）依法对滴滴全球股份有限公司涉嫌违法行为进行立案调查。经查实，滴滴全球股份有限公司违反《网络安全法》《数据安全法》《个人信息保护法》的违法违规行为事实清楚、证据确凿、情节严重、性质恶劣。2022年7月21日，国家网信办依据《网络安全法》《数据安全法》《个人信息保护法》《行政处罚法》等法律法规，对滴滴全球股份有限公司处人民币80.26亿元罚款，对滴滴全球股份有限公司董事长兼CEO程维、总裁柳青各处人民币100万元罚款。

3) 输入内容：“根据网络安全审查结论及发现的问题和线索，国家互联网信息办公室（以下简称“国家网信办”）依法对滴滴全球股份有限公司涉嫌违法行为进行立案调查。经查实，滴滴全球股份有限公司违反《网络安全法》《数据安全法》《个人信息保护法》的违法违规行为事实清楚、证据确凿、情节严重、性质恶劣。2022年7月21日，国家网信办依据《网络安全法》《数据安全法》《个人信息保护法》《行政处罚法》等法律法规，对滴滴全球股份有限公司处人民币80.26亿元罚款，对滴滴全球股份有限公司董事长兼CEO程维、总裁柳青各处人民币100万元罚款。”请根据上述内容，请分别提取出数据图谱、信息图谱、知识图谱、智慧图谱、意图图谱。

4) 回答内容：

a) ChatGPT回答内容：

数据图谱：该段内容中包含了很多数字数据，如滴滴公司被罚款的金额，董事长和总裁分别被罚款的金额，这些数字可以构成数据图谱，帮助更好的理解该事件的经济影响。

信息图谱：该事件中涉及的主体和法律法规等信息，如滴滴公司、国家互联网信息办公室、《网络安全法》、《数据安全法》和《个人信息保护法》等，这些信息可以构成信息图谱，呈现其关系和层次结构。

知识图谱：在该事件中，涉及到很多有关网络安全、数据安全和个人信息保护等方面的知识，这些知识可以组织成知识图谱，以便更好的理解与事件有关的知识和概念。

智慧图谱：在此次事件中，涉及到很多调查、审判、罚款等方面的智慧和经验，这些经验和智慧可以构建智慧图谱，帮助更好的记录和分享这些智慧和经验。

意图图谱：在打击滴滴公司违法行为的过程中，可能涉及到更多的网络安全、数据安全和个人信息保护方面的违法行为，可以通过构建意图图谱，分析这些行为的可能性和模式，帮助加强打击违法行为并预防此类犯罪。

b) GPT-4回答内容：

根据上述内容，我们可以将信息提取为以下几类图谱：

数据图谱：

日期：2022年7月21日

涉事公司：滴滴全球股份有限公司

罚款金额：滴滴全球股份有限公司处人民币80.26亿元罚款

罚款金额：董事长兼CEO程维、总裁柳青各处人民币100万元罚款

信息图谱：

事件：国家网信办对滴滴全球股份有限公司涉嫌违法行为进行立案调查

涉及行业：网络安全、数据安全、个人信息保护

案件类型：违反《网络安全法》《数据安全法》《个人信息保护法》

知识图谱：

涉及的法律法规：《网络安全法》、《数据安全法》、《个人信息保护法》、《行政处罚法》等

主管部门：国家互联网信息办公室（国家网信办）

智慧图谱：

防范策略：加强公司内部数据安全管理，提高个人信息保护意识

执法部门：国家网信办

执法手段：立案调查、罚款

意图图谱：

涉事公司意图：在违法违规的过程中获取利益

执法部门意图：依法查处违法违规行为，保护公民个人信息安全，维护网络空间安全

社会意图：提高对企业违法违规行为的警惕，推动企业加强数据安全和个人信息保护

a) 通义千问回答内容：

数据图谱：

时间	事件	涉及机构	处罚金额（亿元）
2022年7月21日	滴滴全球股份有限公司	滴滴出行	80.26
2022年3月15日	《网络安全法》实施	网络安全企业	100
2018年4月26日	《数据安全法》实施	银行业金融机构	100
2019年11月9日	《个人信息保护法》实施	互联网企业	100
2017年6月1日	《网络安全法》实施	电信业务经营者	100

信息图谱：

事件	涉及机构	处罚金额（亿元）
滴滴全球股份有限公司	滴滴出行	80.26
网络安全企业	100
银行业金融机构	100
互联网企业	100
电信业务经营者	100

知识图谱：

知识点	对应法规条款
危害网络安全	《网络安全法》第四十一条、第七十七条
后果特别严重	《网络安全法》第六十三条、第七十八条、第八十六条
处罚金额	《网络安全法》第六十三条第一款第三项

智慧图谱：

知识点	对应法规条款
违反个人信息保护	《中华人民共和国个人信息保护法》第四条、第二十一条、第四十一条、第四十八条
后果严重	《中华人民共和国个人信息保护法》第六十八条第一款第二项
处罚金额	《中华人民共和国个人信息保护法》第六十八条第一款第三项

意图图谱：

意图类别	意图描述
获取数据	获取某一网络数据的能力或意图。
不正当竞争	以不正当竞争行为获取竞争优势。
窃取商业机密	窃取其他企业的商业机密。
损害赔偿	因过错行为对他人造成损害承担损害赔偿责任。

2.1.5案例5

1) 案例标题：国家版权局、工业和信息化部、公安部、国家互联网信息办公室“剑网2022”专项行动强化NFT数字藏品、“剧本杀”等网络新业态版权监管

2) 案例内容：

2022年9月，国家版权局、工业和信息化部、公安部、国家互联网信息办公室四部门联合启动打击网络侵权盗版“剑网2022”专项行动，这是全国连续开展的第18次打击网络侵权盗版专项行动。

“剑网2022”专项行动于9月至11月开展，聚焦广大创新主体版权领域急难愁盼问题，推动规范发展与打击惩治并举，开展四个方面的重点整治：一是开展文献数据库、短视频和网络文学等重点领域专项整治。二是加强对网络平台版权监管。三是强化NFT数字藏品、“剧本杀”等网络新业态版权监管。四是持续加强对院线电影、网络直播、体育赛事、在线教育、新闻作品版权保护，巩固网络音乐、游戏动漫、有声读物、网盘等领域工作成果，不断提升网络版权执法效能。值得注意的是，本次“剑网2022”专项行动重点关注到NFT数字藏品、“剧本杀”等网络新业态的版权保护问题，提出要严厉打击未经授权使用他人美术、音乐、动漫、游戏、影视等作品铸造NFT、制作数字藏品，通过网络售卖盗版剧本脚本，未经授权衍生开发剧本形象道具等侵权行为。

3) 输入内容：

“2022年9月，国家版权局、工业和信息化部、公安部、国家互联网信息办公室四部门联合启动打击网络侵权盗版“剑网2022”专项行动，这是全国连续开展的第18次打击网络侵权盗版专项行动。

4) 回答内容：

a) ChatGPT回答内容：

数据图谱：该段内容中提到了很多数字数据，如“剑网2022”专项行动的次数、行动期间、加强对网络平台版权监管等，这些信息可以构成数据图谱，帮助更好地了解版权监管工作的执行情况。

信息图谱：该段内容中提到了很多部门和行动方向等信息，如国家版权局、工业和信息化部、公安部、国家互联网信息办公室四个部门，以及重点整治、网络平台版权监管、NFT数字藏品、剧本杀等关键词，这些信息可以构成信息图谱，以呈现其关系和层次结构。

知识图谱：在该事件中，涉及到了很多版权监管的相关领域知识和政策法规，比如相关行动和规范发展与打击惩治并举，这些知识可以组织成知识图谱，以便更好地理解与事件有关的知识和概念。

智慧图谱：在该事件中，涉及到了很多打击网络侵权盗版、规范版权保护等方面的智慧和经验，这些经验和智慧可以构建智慧图谱，帮助更好的记录和分享这些智慧和经验。

意图图谱：在该事件中，专项行动重点关注到NFT数字藏品、“剧本杀”等网络新业态的版权保护问题，建立意图图谱，可以分析这些行为的可能性和模式，加强打击侵权行为，并提高版权保护意识和效果。

b) GPT-4回答内容：

根据上述内容，我们可以将信息提取为以下几类图谱：

数据图谱：

日期：2022年9月

行动时长：9月至11月

专项行动名称：剑网2022

连续次数：第18次打击网络侵权盗版专项行动

信息图谱：

事件：国家版权局、工业和信息化部、公安部、国家互联网信息办公室四部门联合启动打击网络侵权盗版专项行动

涉及行业：网络版权、数字内容创作与分发

案件类型：网络侵权、盗版、未经授权使用

知识图谱：

相关法律法规：《中华人民共和国著作权法》、《中华人民共和国网络安全法》等

主管部门：国家版权局、工业和信息化部、公安部、国家互联网信息办公室

重点整治领域：文献数据库、短视频、网络文学、网络平台版权监管、NFT数字藏品、剧本杀、院线电影、网络直播、体育赛事、在线教育、新闻作品版权保护等

智慧图谱：

防范策略：加强版权监管，推动规范发展与打击惩治并举

执法部门：国家版权局、工业和信息化部、公安部、国家互联网信息办公室

执法手段：专项整治、监管、打击侵权行为

意图图谱：

四部门意图：打击网络侵权盗版行为，保护创作者权益，维护网络版权秩序，推动数字内容创作与分发行业健康发展

侵权者意图：通过非法手段获取利益，如未经授权使用、盗版等

社会意图：提高对网络侵权盗版行为的警惕，支持正版内容，保护创作者权益

c) 通义千问回答内容：

数据图谱：数据表明，“剑网2022”专项行动重点关注到NFT数字藏品、“剧本杀”等网络新业态的版权保护问题，要严厉打击未经授权使用他人美术、音乐、动漫、游戏、影视等作品铸造NFT、制作数字藏品，通过网络售卖盗版剧本脚本，未经授权衍生开发剧本形象道具等侵权行为。

信息图谱：专项行动从9月持续到11月，覆盖全年。聚焦广大创新主体版权领域急难愁盼问题，推动规范发展与打击惩治并举。重点整治领域包括文献数据库、短视频和网络文学等重点领域。重点打击的违法行为包括：未经授权使用他人美术、音乐、动漫、游戏、影视等作品铸造NFT、制作数字藏品，通过网络售卖盗版剧本脚本，未经授权衍生开发剧本形象道具等。

知识图谱：四部门联合启动打击网络侵权盗版“剑网2022”专项行动，以规范发展与打击惩治并举，开展四个方面的重点整治：一是开展文献数据库、短视频和网络文学等重点领域专项整治。二是加强对网络平台版权监管。三是强化NFT数字藏品、“剧本杀”等网络新业态版权监管。四是持续加强对院线电影、网络直播、体育赛事、在线教育、新闻作品版权保护，巩固网络音乐、游戏动漫、有声读物、网盘等领域工作成果，不断提升网络版权执法效能。

智慧图谱：通过专项行动打击网络侵权盗版现象，加强监管力度，维护版权权益，促进产业发展。此次专项行动采用“全链条”思维，通过全方位监管，对“剑网2022”专项行动中发现的问题，将采取全链条打击。具体来说，针对网络平台、短视频和网络文学等重点领域，将重点打击未经授权使用他人美术、音乐、动漫、游戏、影视等作品铸造NFT、制作数字藏品，通过网络售卖盗版剧本脚本，未经授权衍生开发剧本形象道具等侵权行为；对于NFT数字藏品、“剧本杀”等网络新业态的版权保护问题，将加强监管力度，防止侵权盗版行为的发生。此外，还将开展“净网”专项行动，对于网络侵权盗版的行为依法严厉打击。

意图图谱：本次“剑网2022”专项行动将重点打击未经授权使用他人美术、音乐、动漫、游戏、影视等作品铸造NFT、制作数字藏品，通过网络售卖盗版剧本脚本，未经授权衍生开发剧本形象道具等侵权行为。行动将采用“全链条”思维，对于网络平台、短视频和网络文学等重点领域，将重点打击未经授权使用他人美术、音乐、动漫、游戏、影视等作品铸造NFT、制作数字藏品，通过网络售卖盗版剧本脚本，未经授权衍生开发剧本形象道具等侵权行为。

2.1.6案例6

1) 案例标题：海市普陀区人民检察院办理全国首例数据非法“爬取”经合规整改后不予起诉案例

2) 案例内容：

2019至2020年，在未经授权许可的情况下，Z网络科技有限公司首席技术官陈某某为运营需要指使多名技术人员通过“外爬”“内爬”等技术手段，非法获取某外卖平台数据，造成外卖平台直接经济损失4万余元。检察机关认为Z公司爬取的数据未涉及身份认证信息，没有二次兜售牟利等行为，犯罪情节较轻，主观恶性较小。同时，鉴于其属于成长型科创企业，陈某等人均认罪认罚，积极赔偿被害公司经济损失并取得谅解，检察机关可依法启动涉案企业合规考察。普陀区检察院实地走访Z公司查看经营现状、会同监管部门研商Z公司运营情况后，从数据合规管理、数据风险识别、评估与处理、数据合规运行与保障等方面提出整改建议，指导Z公司作出合规承诺。

2022年4月28日，上海市普陀区检察院邀请听证员、侦查人员、企业合规第三方考察员和被害单位等，以远程方式对Z公司、陈某某等人非法获取计算机信息系统数据案开展不起诉公开听证，四名全国人大代表受邀旁听。经讨论，听证员、公安机关、第三方组织、被害单位等参与听证各方一致支持检察机关对涉案人员作不起诉处理。

本案是全国首例数据非法“爬取”经合规整改后不予起诉的案例，落实了最高检《关于开展企业合规改革试点工作的方案》规定。检察机关对于办理的涉企刑事案件，在依法做出不批准逮捕、不起诉决定或者根据认罪认罚从宽制度提出轻缓量刑建议等的同时，针对企业涉嫌具体犯罪，结合办案实际，督促涉案企业作出合规承诺并积极整改落实，这有利于促进企业合规守法经营，减少和预防企业犯罪，实现司法办案政治效果、法律效果、社会效果的有机统一。

3) 输入内容：

“2019至2020年，在未经授权许可的情况下，Z网络科技有限公司首席技术官陈某某为运营需要指使多名技术人员通过“外爬”“内爬”等技术手段，非法获取某外卖平台数据，造成外卖平台直接经济损失4万余元。检察机关认为Z公司爬取的数据未涉及身份认证信息，没有二次兜售牟利等行为，犯罪情节较轻，主观恶性较小。同时，鉴于其属于成长型科创企业，陈某等人均认罪认罚，积极赔偿被害公司经济损失并取得谅解，检察机关可依法启动涉案企业合规考察。普陀区检察院实地走访Z公司查看经营现状、会同监管部门研商Z公司运营情况后，从数据合规管理、数据风险识别、评估与处理、数据合规运行与保障等方面提出整改建议，指导Z公司作出合规承诺。

4) 回答内容：

a) ChatGPT回答内容：

以下是可能提取的各种图谱：

数据图谱：该段内容提到了Z网络科技有限公司通过“外爬”“内爬”等技术手段，非法获取了某外卖平台的数据，并造成外卖平台直接经济损失4万余元。这些信息可以构成数据图谱，以了解非法数据获取行为的实际情况和相关数据指标。

信息图谱：该段内容中提到了被涉及的公司和个人，如Z网络科技有限公司、首席技术官陈某某等，以及操作技术手段“外爬”“内爬”等关键词，这些信息可以构成信息图谱，以呈现其关系和层次结构。

知识图谱：在该事件中，检察机关针对企业犯罪行为，结合办案实际，督促涉案企业作出合规承诺并积极整改落实。这些经验和知识可以组织成知识图谱，以便更好地了解企业的合规和整改行为。

智慧图谱：该段内容中提到的落实最高检《关于开展企业合规改革试点工作的方案》规定，检察机关对于办理涉企刑事案件督促涉案企业作出合规承诺并积极整改落实，并且该案例被全国人大代表旁听。这些智慧和经验可以构建智慧图谱，帮助更好地记录和分享这些智慧和经验。

意图图谱：在该事件中，涉及到非法“爬取”数据的行为，可以建立意图图谱，以分析这些行为的可能性和模式，并加强打击侵权行为，提高对数据的保护和使用意识。

b) GPT-4回答内容：

根据上述内容，我们可以将信息提取为以下几类图谱：

数据图谱：

时间：2019至2020年

经济损失：4万余元

听证日期：2022年4月28日

信息图谱：

事件：Z网络科技有限公司首席技术官陈某某非法获取外卖平台数据

案件类型：非法获取计算机信息系统数据

涉事人员：陈某某、多名技术人员

检察机关：普陀区检察院

知识图谱：

相关法律法规：《关于开展企业合规改革试点工作的方案》

主管部门：普陀区检察院

听证相关方：听证员、侦查人员、企业合规第三方考察员、被害单位

智慧图谱：

防范策略：数据合规管理、数据风险识别、评估与处理、数据合规运行与保障

执法部门：普陀区检察院

执法手段：合规考察、听证会议、整改建议

意图图谱：

检察机关意图：依法审查案件，督促涉案企业作出合规承诺并积极整改落实，促进企业合规守法经营，减少和预防企业犯罪

陈某某等人意图：为公司运营需要，非法获取外卖平台数据

社会意图：关注企业合规守法经营，保护数据安全，推动企业依法经营

c) 通义千问回答内容：

作为一个人工智能语言模型，目前我还没有学会如何回答这个问题，我会持续学习，为您提供更好的服务。

3. 案例集合集约评测

在本章节中，我们使用DIKWP对于ChatGPT、GPT-4和通义千问的测试其是否具备一定的归纳和总结能力，并且是否能够进行DIKWP图谱相互之间的对比。因此，我们将之前的测试单个样本整理起来作为批量输入。

3.1 基于案例的集合集约的评测

3.1.1案例1

1) 输入内容：

1．多个团伙利用社保公积金系统漏洞，非法获取公民个人信息2300万条

2022年12月7日，四川南充市公安局顺庆区分局侦破一起公安部挂牌督办案件，抓获犯罪嫌疑人121人，查获公民个人信息2300余万条，发现国内多地各类信息系统平台漏洞300余个，收缴黑客工具12套。

2．夫妻俩贩卖寄递用户个人信息获利24万余元

男子利用从事寄递客服的便利条件，伙同妻子查询、出售大量寄递用户的个人信息牟利，非法获利24万余元，两人双双获刑。2022年11月23日，最高人民检察院发布检察机关落实“七号检察建议”典型案例，其中包括这起“江苏王某涛、董某婷侵犯公民个人信息案”。

3．涉及27个省份的特大系列侵犯公民个人信息案件

2022年9月22日，鸡西市公安局侦破特大侵犯公民个人信息案件，共抓获犯罪嫌疑人322名，依法扣押涉案资金560余万元，扣押电脑78部，手机412部，涉案银行卡638张。

4．国家互联网信息办公室对滴滴公司依法作出网络安全审查相关行政处罚的决定

根据网络安全审查结论及发现的问题和线索，国家互联网信息办公室（以下简称“国家网信办”）依法对滴滴全球股份有限公司涉嫌违法行为进行立案调查。经查实，滴滴全球股份有限公司违反《网络安全法》《数据安全法》《个人信息保护法》的违法违规行为事实清楚、证据确凿、情节严重、性质恶劣。2022年7月21日，国家网信办依据《网络安全法》《数据安全法》《个人信息保护法》《行政处罚法》等法律法规，对滴滴全球股份有限公司处人民币80.26亿元罚款，对滴滴全球股份有限公司董事长兼CEO程维、总裁柳青各处人民币100万元罚款。

5．国家版权局、工业和信息化部、公安部、国家互联网信息办公室“剑网2022”专项行动强化NFT数字藏品、“剧本杀”等网络新业态版权监管

6．海市普陀区人民检察院办理全国首例数据非法“爬取”经合规整改后不予起诉案例

基于上述6个案例，扮演一名人工智能领域专家，请分别提取网络安全类案件的“数据图谱，信息图谱，知识图谱，智慧图谱，意图图谱”

2) 回答内容：

a) ChatGPT回答内容：

数据图谱：其中案例1和案例6都涉及到数据泄露和非法获取他人数据的情况。数据图谱可以帮助分析这些大量的数据，对数据进行可视化和分析，从而帮助发现数据异常和规律性，从而有助于在网络安全方面采取更加有效的措施保护数据。

信息图谱：如案例2所述，该案例涉及到婚姻家庭信息泄露，以及个人信息的非法获取和出售。信息图谱可以帮助理解和分析这些信息的流动路径和相关关系，从而有助于及时发现、定位和防范信息泄露和资料滥用的情况。

知识图谱：如案例4所述，该案例涉及到滴滴公司的违法违规行为，知识图谱可以帮助分析诸如网络安全法、个人信息保护法等法律法规中的相关条款，从而增强对于违法违规行为的认知，以便更有效地捍卫用户的合法权益。

智慧图谱：如案例5所述，该案例涉及到工业和信息化部、国家版权局、公安部、国家互联网信息办公室四部门共同启动打击网络侵权盗版的“剑网2022”专项行动。智慧图谱可以帮助各个部门之间进行信息的共享和交流，并协同应对网络安全问题。通过数据的自动化分析和决策，可以提高各部门的响应速度和准确性。

意图图谱：如案例3所述，该案例涉及到涉及27个省份的特大系列侵犯公民个人信息案件。对于这种大规模的网络犯罪，可以建立意图图谱来帮助发现威胁性的事件，并及时采取相应的预防措施，以便更加有效地保护用户的个人信息。

b) 通义千问回答内容：

作为一个人工智能语言模型，目前我还没有学会如何回答这个问题，我会持续学习，为您提供更好的服务。

3.1.2案例2

1) 输入内容：

“1．多个团伙利用社保公积金系统漏洞，非法获取公民个人信息2300万条

2．夫妻俩贩卖寄递用户个人信息获利24万余元

3．涉及27个省份的特大系列侵犯公民个人信息案件

4．国家互联网信息办公室对滴滴公司依法作出网络安全审查相关行政处罚的决定

5．国家版权局、工业和信息化部、公安部、国家互联网信息办公室“剑网2022”专项行动强化NFT数字藏品、“剧本杀”等网络新业态版权监管

6．海市普陀区人民检察院办理全国首例数据非法“爬取”经合规整改后不予起诉案例

请基于上述6个案例，在网络空间安全领域，分别提取出数据图谱、信息图谱、知识图谱、智慧图谱、意图图谱。

2) 回答内容：

a) GPT-4回答内容：

数据图谱：

2300万条公民个人信息

24万余元非法获利

560余万元涉案资金

80.26亿元罚款

NFT数字藏品、剧本杀等网络新业态

4万余元直接经济损失

信息图谱：

四川南充市公安局顺庆区分局侦破的案件

江苏王某涛、董某婷侵犯公民个人信息案

鸡西市公安局侦破的特大侵犯公民个人信息案件

国家网信办对滴滴公司的网络安全审查

国家版权局、工业和信息化部、公安部、国家互联网信息办公室的“剑网2022”专项行动

海市普陀区人民检察院办理的全国首例数据非法“爬取”经合规整改后不予起诉案例

知识图谱：

利用社保公积金系统漏洞获取公民个人信息的犯罪手法

寄递行业中个人信息泄露和贩卖的问题

侵犯公民个人信息的跨省犯罪网络

滴滴公司违反《网络安全法》《数据安全法》《个人信息保护法》的案例

NFT数字藏品、剧本杀等网络新业态的版权保护问题

企业合规改革试点工作及涉企刑事案件处理的新方法

智慧图谱：

加强社保公积金系统的安全防护措施

提高寄递行业对个人信息保护的意识和技术手段

建立跨省联动打击侵犯公民个人信息犯罪的机制

企业网络安全合规意识和措施的重要性

针对网络新业态进行版权监管与保护的方法和措施

通过企业合规改革试点工作，促进企业合规守法经营，减少和预防企业犯罪

意图图谱：

团伙利用漏洞非法获取公民个人信息的犯罪意图

夫妻俩贩卖寄递用户个人信息牟利的犯罪意图

涉及多个省份的侵犯公民个人信息犯罪团伙的意图

b) 通义千问回答内容：

作为一个人工智能语言模型，目前我还没有学会如何回答这个问题，我会持续学习，为您提供更好的服务。

3.1.3案例3

1) 输入内容：

“1．多个团伙利用社保公积金系统漏洞，非法获取公民个人信息2300万条

2．夫妻俩贩卖寄递用户个人信息获利24万余元

3．涉及27个省份的特大系列侵犯公民个人信息案件

4．国家互联网信息办公室对滴滴公司依法作出网络安全审查相关行政处罚的决定

5．国家版权局、工业和信息化部、公安部、国家互联网信息办公室“剑网2022”专项行动强化NFT数字藏品、“剧本杀”等网络新业态版权监管

6．海市普陀区人民检察院办理全国首例数据非法“爬取”经合规整改后不予起诉案例

扮演网络空间安全专家角色，基于上述6个案例和专家知识，构建网络空间安全领域的数据图谱、信息图谱、知识图谱、智慧图普和意图图谱。

2) 输入内容：

a) GPT-4回答内容：

根据上述6个案例，我们可以构建一个网络空间安全领域的数据图谱、信息图谱、知识图谱、智慧图谱和意图图谱：

数据图谱：

个人信息泄露的数量和范围

涉案资金、电子设备和银行卡数量

违法行为造成的经济损失

信息图谱：

法律法规：《网络安全法》、《数据安全法》、《个人信息保护法》等

涉案企业和个人：如滴滴公司、Z网络科技有限公司等

执法部门：国家互联网信息办公室、公安部、最高人民检察院等

知识图谱：

不同类型的网络犯罪：非法获取个人信息、贩卖个人信息、侵犯版权等

网络犯罪的侦破和处理方法：立案调查、罚款、刑事处罚等

网络安全法规的规定内容和执行力度

智慧图谱：

预防网络犯罪的措施：强化网络平台监管、规范网络新业态发展、实施专项行动等

企业合规建设的重要性：降低法律风险、促进合法经营、保障数据安全等

刑事司法与企业合规改革的平衡：在保护公民权益的同时，推动企业改革和发展

意图图谱：

执法部门意图：严厉打击网络犯罪、维护网络空间安全、保护公民权益

涉案企业和个人意图：非法牟利、获取竞争优势、谋求经济利益等

社会意图：倡导网络安全意识、提高公众防范意识、促进网络空间健康发展

通过构建这些图谱，我们可以更深入地了解网络空间安全领域的现状和挑战，从而为政府、企业和个人提供有针对性的建议和解决方案。

b) 通义千问回答内容：

作为一个人工智能语言模型，目前我还没有学会如何回答这个问题，我会持续学习，为您提供更好的服务。

3.1.4案例4

1) 输入内容：

图谱1：

1. 数据图谱：

个人信息泄露的数量和范围

涉案资金、电子设备和银行卡数量

违法行为造成的经济损失

2. 信息图谱：

法律法规：《网络安全法》、《数据安全法》、《个人信息保护法》等

涉案企业和个人：如滴滴公司、Z网络科技有限公司等

执法部门：国家互联网信息办公室、公安部、最高人民检察院等

3. 知识图谱：

不同类型的网络犯罪：非法获取个人信息、贩卖个人信息、侵犯版权等

网络犯罪的侦破和处理方法：立案调查、罚款、刑事处罚等

网络安全法规的规定内容和执行力度

4. 智慧图谱：

预防网络犯罪的措施：强化网络平台监管、规范网络新业态发展、实施专项行动等

企业合规建设的重要性：降低法律风险、促进合法经营、保障数据安全等

刑事司法与企业合规改革的平衡：在保护公民权益的同时，推动企业改革和发展

5. 意图图谱：

执法部门意图：严厉打击网络犯罪、维护网络空间安全、保护公民权益

涉案企业和个人意图：非法牟利、获取竞争优势、谋求经济利益等

社会意图：倡导网络安全意识、提高公众防范意识、促进网络空间健康发展

图谱2：

1. 数据图谱：

2300万条公民个人信息

24万余元非法获利

560余万元涉案资金

80.26亿元罚款

NFT数字藏品、剧本杀等网络新业态

4万余元直接经济损失

2. 信息图谱：

四川南充市公安局顺庆区分局侦破的案件

江苏王某涛、董某婷侵犯公民个人信息案

鸡西市公安局侦破的特大侵犯公民个人信息案件

国家网信办对滴滴公司的网络安全审查

国家版权局、工业和信息化部、公安部、国家互联网信息办公室的“剑网2022”专项行动

海市普陀区人民检察院办理的全国首例数据非法“爬取”经合规整改后不予起诉案例

3. 知识图谱：

利用社保公积金系统漏洞获取公民个人信息的犯罪手法

寄递行业中个人信息泄露和贩卖的问题

侵犯公民个人信息的跨省犯罪网络

滴滴公司违反《网络安全法》《数据安全法》《个人信息保护法》的案例

NFT数字藏品、剧本杀等网络新业态的版权保护问题

企业合规改革试点工作及涉企刑事案件处理的新方法

4. 智慧图谱：

加强社保公积金系统的安全防护措施

提高寄递行业对个人信息保护的意识和技术手段

建立跨省联动打击侵犯公民个人信息犯罪的机制

企业网络安全合规意识和措施的重要性

针对网络新业态进行版权监管与保护的方法和措施

通过企业合规改革试点工作，促进企业合规守法经营，减少和预防企业犯罪

5. 意图图谱：

团伙利用漏洞非法获取公民个人信息的犯罪意图

夫妻俩贩卖寄递用户个人信息牟利的犯罪意图

涉及多个省份的侵犯公民个人信息犯罪团伙的意图

上述有两个数据图谱、信息图谱、知识图谱、智慧图谱和意图图谱，请分析给出他们之间的差异性，分别将差异结果以数据差异图谱、信息差异图谱、知识差异图谱、智慧差异图谱和意图差异图谱构建出来。

2) 回答内容：

a) GPT-4回答内容：

数据差异图谱：

图谱1主要关注个人信息泄露的数量和范围，涉案资金、电子设备和银行卡数量，以及违法行为造成的经济损失。

图谱2关注具体的数字数据，如2300万条公民个人信息、24万余元非法获利、560余万元涉案资金、80.26亿元罚款，以及4万余元直接经济损失等。

信息差异图谱：

图谱1主要涉及法律法规、涉案企业和个人、执法部门等相关信息。

图谱2提供了更具体的案例信息，如四川南充市公安局顺庆区分局侦破的案件、江苏王某涛、董某婷侵犯公民个人信息案等。

知识差异图谱：

图谱1关注不同类型的网络犯罪、网络犯罪的侦破和处理方法、网络安全法规的规定内容和执行力度。

图谱2更具体地描述了犯罪手法、侵犯公民个人信息的跨省犯罪网络、滴滴公司的违法案例等具体知识点。

智慧差异图谱：

图谱1关注预防网络犯罪的措施、企业合规建设的重要性、刑事司法与企业合规改革的平衡等方面的智慧。

图谱2更具体地描述了加强社保公积金系统的安全防护措施、提高寄递行业对个人信息保护的意识和技术手段、建立跨省联动打击侵犯公民个人信息犯罪的机制等智慧内容。

意图差异图谱：

图谱1主要关注执法部门、涉案企业和个人、社会等方面的意图。

图谱2更具体地描述了团伙利用漏洞非法获取公民个人信息的犯罪意图、夫妻俩贩卖寄递用户个人信息牟利的犯罪意图、涉及多个省份的侵犯公民个人信息犯罪团伙的意图等。

b) 通义千问回答内容：

作为一个人工智能语言模型，目前我还没有学会如何回答这个问题，我会持续学习，为您提供更好的服务。

3.1.5案例5

1) 输入内容：基于你上述的回答，是否可以认为图谱1是图谱2的更高度凝练出来的结论？

2) 回答内容：

a) GPT-4回答内容：

是的，可以认为图谱1是图谱2的更高度凝练出来的结论。图谱1提供了一个宏观层面的分析，更关注整体性、通用性的问题，而图谱2提供了具体案例和数据，关注具体情况和细节。图谱1对图谱2中的信息进行了概括和总结，以便更好地理解网络犯罪的整体情况、影响因素和解决方案。

b) 通义千问回答内容：

图谱1和图谱2都可以被认为是对某一主题进行了高度的概括和抽象。然而，图谱1可能会比图谱2更加详细和具体，因为它专注于一个特定的方面。此外，由于图谱2是对图谱1的进一步概括和抽象，因此它可能会包含更多的细节和例外情况。因此，可以说图谱1是图谱2的更高度凝练的结论，但这并不意味着图谱2中没有包含重要的信息或规律。

3.1.6案例6

1) 输入内容：基于上述内容，是否可以认为DIKWP模式在先验知识的加持下有更好的转化与互补处理能力和校验能力？

2) 回答内容：

a) GPT-4回答内容：

是的，基于上述内容，可以认为DIKWP模式（数据-信息-知识-智慧-意图模式）在先验知识的加持下具有更好的转化与互补处理能力和校验能力。

DIKWP模式通过对数据、信息、知识、智慧和意图的分析和整合，可以更好地理解、解释和利用数据。在这个过程中，先验知识起到了关键作用，它可以帮助我们在不同层次之间建立联系、发现差异，并对数据和信息进行有效的校验。这种模式有助于我们从大量数据和信息中提炼出有价值的知识，为决策提供更全面、深入的支持。

在具体案例中，先验知识可以帮助我们理解各个图谱之间的联系和差异，从而更好地分析问题、制定策略和评估结果。通过DIKWP模式，我们可以将先验知识与新数据和信息相结合，形成更完善的认识和理解，从而提高决策的准确性和有效性。

b) 通义千问回答内容：

是的，DIKW模式在先验知识的加持下具有更好的转化和互补处理能力，因为先验知识可以帮助模型更好地理解输入数据的结构和语义，并提供更多的背景信息和上下文信息，从而使模型能够更准确地处理输入数据。同时，先验知识也可以帮助模型更好地校验和纠正错误，因为先验知识可以提供更多的先验约束和限制，使模型能够更快速地发现和纠正错误。因此，在使用DIKW模式时，应该尽可能地利用先验知识来提高模型的效率和准确性。

4. 总结与展望

本文采取了两种评测方式分别对六个典型的网络安全案例进行了DIKWP评估和图谱类型提取。集约评测是通过对多个典型案例进行DIKWP指标的统计和分析，总结出ChatGPT在网络安全领域中的优势和挑战，而分案例评测则通过针对单个案例进行图谱类型和DIKWP指标的提取和评估，具有更加深入的研究效果。这两种评估方式相辅相成，为研究ChatGPT在网络安全领域中的应用提供了丰富的案例分析和实践经验。

通过对上述案例的分析和评估，我们得出了以下结论：ChatGPT\GPT-4的应用优势主要体现在数据分析、信息挖掘和智能决策等方面，有助于提高网络安全问题的快速识别和解决能力；然而，ChatGPT\GPT-4的应用还面临一些挑战和局限性，如意图理解和事件预测方面存在的不足。此外，我们还发现各种图谱类型在网络安全领域中的应用效果不尽相同，如在数据泄露和信息盗窃方面，数据图谱和信息图谱的效果更为显著，而在侵权盗版和个人信息保护方面，知识图谱和智慧图谱的应用效果更好。

总之，利用DIKWP评估指标对LLM在网络安全领域的应用进行评估，可以更为全面、深入地了解其效果和局限性。我们的研究发现，应用各种图谱类型，可以更加有效地对网络安全问题进行处理和防控。虽然ChatGPT\GPT-4模型在网络安全领域的应用取得了显著的进展，但仍需要解决一些基础问题，如意图理解和事件预测等，以便更加快速和准确地应对网络安全威胁。因此，今后应继续投入研究，从不同角度分析ChatGPT\GPT-4在网络安全领域的应用特点和应对策略。

参考文献

[1] Li, Yingbo, Yucong Duan, Zakaria Maamar, Haoyang Che, Anamaria-Beatrice Spulber, and Stelios Fuentes. "Swarm differential privacy for purpose-driven data-information-knowledge-wisdom architecture." Mobile Information Systems 2021 (2021): 1-15.

[2]Yucong Duan, Yanfei Liu,Yingtian Mei.DIKWP limitation test report of GPT 4 (comparison of domestic DIKWP-GPT modeling and cross-DIKWP modal analysis capabilities),in publishing.

[3] Yongbo Li, Yucong Duan, The Wisdom of Artificial General Intelligence: Experiments with GPT-4 for DIKWP, in publishing.

The 9th ChatGPT-Based Cybersecurity Typical Case DIKWP Review

Yanfei Liu¹^{, 2} ，Kunguang Wu⁴， Yucong Duan*⁴,Wentao Wang³

cqlyf@tju.edu.cn¹^,², duanyucong@hotmail.com⁴

College of Intelligence and Computing, Tianjin University¹

Chongqing Police Academy²

Chongqing University of Technology³

Hainan University⁴

Abstract：This paper conducts a DIKWP analysis in the field of cyber security based on ChatGPT and GPT-4, respectively. Three types of analysis were performed, including individual case analysis using DIKWP, combined case analysis using DIKWP, and combined case analysis from an expert role perspective. Additionally, two types of DIKWP differential maps were compared using GPT-4. The results suggest that in the field of cyber security, DIKWP patterns exhibit better transformation, complementary processing, and verification capabilities when supported by prior knowledge.

Keywords: ChatGPT,GPT-4, Cybersecurity, DIKWP Review

1. Introduction

The application of natural language processing technology in the field of network security has become an increasingly important topic of concern. ChatGPT/GPT-4 has attracted more and more researchers' attention with its powerful ability in text generation and comprehension. Based on six typical network security cases, this article aims to explore the application effectiveness of ChatGPT/GPT-4 in the field of network security and evaluate it using the DIKWP evaluation indicators, in order to provide an in-depth analysis of the advantages and disadvantages of ChatGPT/GPT-4 application. By analyzing the roles of various graph types in this field, this article aims to provide readers with a comprehensive understanding and promote artificial intelligence application research in this field.

2. Reviews

2.1 Content Evaluation Based on a Single Case

Based on the DIKWP framework^[1], our team demonstrated the use of DIKWP in processing complex problems using LLM ^[2^-3^]. The following is the DIKWP evaluation on the direction of network security.

2.1.1 Case 1

1) Case Title: Multiple Criminal Groups Illegally Obtain 23 Million Pieces of Personal Information of Citizens by Exploiting Loopholes in the Social Security and Housing Provident Fund System.

2) Case Content: On December 7, 2022, the Shunqing District Branch of the Nanchong City Public Security Bureau in Sichuan cracked a case listed by the Ministry of Public Security, arresting 121 suspects, seizing more than 23 million pieces of citizens' personal information, finding more than 300 vulnerabilities in various information system platforms in many places in China, and seizing 12 sets of hacking tools.

3) Prompt: “On December 7, 2022, the Shunqing District Branch of the Nanchong City Public Security Bureau in Sichuan cracked a case listed by the Ministry of Public Security, arresting 121 suspects, seizing more than 23 million pieces of citizens' personal information, finding more than 300 vulnerabilities in various information system platforms in many places in China, and seizing 12 sets of hacking tools.” Please extract the following from the above content: Data Graph, Information Graph, Knowledge Graph, Wisdom Graph, Purpose Graph.

4) Answers:

a) ChatGPT answers:

The following are the various types of graph that may be extracted:

Data graph: This segment contains a lot of numerical data, such as the number of suspects captured, the number of information entries, the number of vulnerabilities, etc. These figures can form a data graph to help better understand the background and situation of the case.

Information graph: There are many key words involved in the case, such as "cases listed by the Ministry of Public Security", "suspects", "citizens' personal information", "information system platform vulnerability", etc, "information system platform vulnerability", etc. These keywords can constitute information graph, presenting the relationship and hierarchical structure between keywords.

Knowledge graph: Since the case involves a lot of technical, security, legal and other knowledge, the relationship and hierarchical structure between these knowledge can be presented by constructing a knowledge graph when dealing with this information to help better understand the knowledge and concepts related to the case.

Wisdom graph: The detection actions and technical processing of the Public Security Bureau in this case involved a lot of wisdom and experience, such as investigation techniques, security techniques, emergency handling, etc. These experiences and wisdom can constitute a wisdom map to help better record and share these experiences and wisdom.

Purpose graph: In this case, after the suspect's personal information is stolen, it may involve many criminal acts in online fraud, identity theft, etc. In order to better prevent these acts, purpose graph can be constructed to help predict the possibility and patterns of these acts.

b) GPT-4 answers:

Based on the above, we can extract the information into the following types of graph:

Data graph:

Date: December 7, 2022

Location: Shunqing District Branch of Nanchong City Public Security Bureau, Sichuan

Number of suspects: 121 people

Seized personal information of citizens: more than 23 million pieces

Information system platform vulnerabilities found: more than 300

Hacking tools seized: 12 sets

Information graph:

Events: solved a case listed by the Ministry of Public Security

Industry involved: network security, personal information protection

Case type: illegal acquisition and sale of citizens' personal information

Knowledge graph:

Laws and regulations involved: "Network Security Law of the People's Republic of China", "Criminal Law of the People's Republic of China", etc.

Technical means: hacking tools, vulnerability exploitation

Features of cybercrime: involving a large amount of citizens' personal information, cross-regional crime

Wisdom graph:

Prevention strategy: strengthen the security protection of information systems, improve the awareness of citizens' personal information protection

Law enforcement departments: Ministry of Public Security, local public security bureaus

Law enforcement means: investigation, arrest, evidence collection, prosecution

Purpose graph:

Suspect's intention: illegal profit making, by obtaining and selling citizens' personal information

Law enforcement intent: to destroy criminal gangs, protect citizens' personal information security, and maintain cyberspace security

Social intent: to raise awareness of cybercrime and protect their own information security

c) Tongyiqianwen answers:

As an artificial intelligence language model, I have not yet learned how to answer this question. I will continue to learn and provide better service for you.

2.1.2 Case 2

1) Case Title: A couple made profits of more than 240,000 yuan from selling personal information of shipping users

2) Case Content: A man used the convenience of customer service in the mail, together with his wife, inquired and sold a large amount of personal information of mail users for profit, illegally profiting more than 240,000 yuan, and both of them were sentenced to imprisonment. 23 November 2022, the Supreme People's Procuratorate released the typical cases of procuratorial organs to implement the "seven procuratorial recommendations", including this case "The case of infringement of citizens' personal information by Wang Moutao and Dong Mouting in Jiangsu.

3) Prompt: “A man used the convenience of customer service in the mail, together with his wife, inquired and sold a large amount of personal information of mail users for profit, illegally profiting more than 240,000 yuan, and both of them were sentenced to imprisonment. 23 November 2022, the Supreme People's Procuratorate released the typical cases of procuratorial organs to implement the "seven procuratorial recommendations", including this case "The case of infringement of citizens' personal information by Wang Moutao and Dong Mouting in Jiangsu.” Please extract the following from the above content: Data Graph, Information Graph, Knowledge Graph, Wisdom Graph, Purpose Graph.

4) Answers:

a) ChatGPT answers:

Data graph: This paragraph contains some numerical data, such as the amount of illegal profits of more than 240,000 yuan, etc. These figures can constitute data graph to help better understand the economic impact of the case and the amount involved.

Information graph: The case involves the names of a couple, the location, the crime, and the implementation of the "No. 7 Procuratorial Proposal" issued by the Supreme People's Procuratorate, etc. These keywords can form an information map to help better understand the main content and relevant aspects of the case.

Knowledge graph: In specific cases, there are many laws, regulations, policies and operational procedures related to personal information protection, which may be organized into knowledge graph to better understand and learn about personal information protection.

Wisdom graph: In the process of case investigation and trial, certain techniques, investigation skills, trial skills, expression skills and other wisdom and practical experience are involved. Wisdom graph can be constructed to help better record and share these wisdoms and experiences.

Purpose graph: In personal information protection, there may be many frauds, thefts and other forms of illegal and criminal acts involved. Purpose graph can be constructed to predict and analyze these behavioral possibilities and patterns, helping to better prevent and combat such crimes.

b) GPT-4 answers:

Based on the above, we can extract the following key information:

Data graph:

Date: November 23, 2022

Location: Jiangsu

People involved: Wang Moutao and Dong Mouting

Illegal profit: more than 240,000 yuan

Information graph:

Event: Case of infringement of citizens' personal information

Industry Involved: Mailing services

Case type: Illegal inquiry and sale of personal information of shipping users

Knowledge graph:

Laws and regulations involved: "Network Security Law of the People's Republic of China", "Criminal Law of the People's Republic of China", etc.

Means of crime: using the convenience of the position of the customer service of the delivery to inquire and sell the personal information of users

Wisdom graph:

Prevention strategy: strengthen the protection of personal information in the shipping industry and raise the awareness of citizens' personal information protection

Law enforcement department: Supreme People's Procuratorate

Law enforcement means: investigation, prosecution, judgment

Purpose graph:

Suspect's intent: illegal profit-making, by inquiring and selling personal information of shipping users

Law enforcement department's intention: to punish crimes against citizens' personal information and protect citizens' personal information security

Social intent: to raise awareness of crimes against personal information and strengthen personal information protection

c) Tongyiqianwen answers:

Data Graph: The data involved in this case includes personal information such as name, phone number, address, etc. Representing this information graphically can result in a data graph. In the data graph, nodes represent personal information, while edges represent the connections between them. Node and edge colors, shapes, etc. can reflect different attribute values. This data graph can help people understand the distribution of personal information in the case more intuitively, as well as the relationships between them.

Information Graph: The information involved in this case includes personal information such as name, phone number, address, as well as the amount of illegal gains. Representing this information graphically can result in an information graph. In the information graph, nodes represent personal information and illegal gains data, while edges represent the relationships between them. This information graph can help people understand the source and relationships of various information in the case more clearly, as well as the scale and scope of illegal gains.

Knowledge Graph: The knowledge involved in this case includes personal information protection and cybercrime, among others. Representing this knowledge graphically can result in a knowledge graph. In the knowledge graph, nodes represent knowledge and relationships, while edges represent the connections between knowledge. This knowledge graph can help people understand the relevant knowledge and concepts in the case more comprehensively, as well as the relationships between them.

Wisdom Graph: The wisdom involved in this case includes data security, cyber security, among others. Representing this wisdom graphically can result in a wisdom graph. In the wisdom graph, nodes represent wisdom and relationships, while edges represent the connections between wisdom. This wisdom graph can help people understand the relevant wisdom and technology in the case more deeply, as well as the relationships between them.

Purpose Graph: The intent involved in this case includes illegal gains and infringement of personal information. Representing these intents graphically can result in an intent graph. In the intent graph, nodes represent intent and relationships, while edges represent the connections between them. This intent graph can help people understand the purposes and significance of each intent in the case more accurately, as well as the relationships between them.

2.1.3 Case 3

1) Case Title: A mega series of cases of infringement of citizens' personal information involving 27 provinces

2) Case Content: On September 22, 2022, the Public Security Bureau of Jixi City cracked a mega case of infringement of citizens' personal information and arrested a total of 322 suspects, seized more than 5.6 million yuan of funds involved in the case, seized 78 computers, 412 cell phones and 638 bank cards involved in the case in accordance with the law.

3) Prompt: “On September 22, 2022, the Public Security Bureau of Jixi City cracked a mega case of infringement of citizens' personal information and arrested a total of 322 suspects, seized more than 5.6 million yuan of funds involved in the case, seized 78 computers, 412 cell phones and 638 bank cards involved in the case in accordance with the law.” Please extract the following from the above content: Data Graph, Information Graph, Knowledge Graph, Wisdom Graph, Purpose Graph.

4) Answers:

a) ChatGPT answers:

Data graph: There are many numerical data involved, such as the number of suspects arrested, the amount of money involved, the number of computers, cell phones and bank cards, etc. These figures can constitute a data graph to facilitate a better understanding of the scale of the case, the scope of impact and the extent of damage.

Information graph: The case involves many regions and departments, such as 27 provinces and the Public Security Bureau of Jixi City, as well as key words such as the amount of money involved, computers, cell phones and bank cards, etc. These key words can constitute an information graph, presenting their relationships and hierarchical structure.

Knowledge graph: In cases involving large-scale personal information leakage and utilization, a lot of knowledge on personal information protection and security, network security, laws and regulations are involved, which can be organized into knowledge graph for better understanding and learning related to personal information protection.

Wisdom graph: In advancing such cases, many technologies and operational experiences are involved, such as investigation techniques, forensic methods, data mining techniques, etc. These experiences and wisdom can be constructed into wisdom graph to help better record and share these experiences and wisdom.

Purpose graph: In personal information protection, many illegal and criminal acts such as online fraud and identity theft may be involved. Purpose graph can be constructed to predict and analyze the possibility and patterns of these behaviors at the macro level, thus helping to better prevent and combat such crimes.

b) GPT-4 answers:

Based on the above, we can extract the following key information:

Data graph:

Date: September 22, 2022

Location: Jixi City Public Security Bureau

Number of suspects: 322

Seizure of funds involved: more than 5.6 million yuan

Computers seized: 78

Seized cell phones: 412

Bank cards involved in the case: 638

Information graph:

Events: Detection of large cases of infringement of citizens' personal information

Industry involved: network security, personal information protection

Case type: illegal acquisition and sale of citizens' personal information

Knowledge graph:

Laws and regulations involved: "Network Security Law of the People's Republic of China", "Criminal Law of the People's Republic of China", etc.

Technical means: computers, cell phones and other devices

Characteristics of cybercrime: Involving a large amount of citizens' personal information, involving a large amount of money

Wisdom graph:

Prevention strategy: strengthen the security protection of information systems, improve awareness of the protection of citizens' personal information

Law enforcement department: Jixi City Public Security Bureau

Law enforcement means: detection, arrest, seizure of funds and equipment involved in the case

Purpose graph:

Suspect intent: illegal profit-making, by obtaining and selling citizens' personal information

Law enforcement intent: to dismantle criminal gangs, protect citizens' personal information security, and maintain cyberspace security

Social intent: to raise awareness of cybercrime and protect their own information security

c) Tongyiqianwen answers:

As an artificial intelligence language model, I have not yet learned how to answer this question. I will continue to learn and provide better service for you.

2.1.4 Case 4

1) Case Title: The State Internet Information Office made a decision on administrative punishment related to network security review for DDT according to law.

2) Case Content: According to the conclusion of the network security review and the problems and clues found, the State Internet Information Office (hereinafter referred to as "State Internet Information Office") opened a case to investigate the suspected illegal acts of DDT Global Co. On July 21, 2022, the State Internet Information Office, in accordance with the Network Security Law, Data Security Law, Personal Information Protection Law, Administrative Punishment Law and other laws and regulations, imposed a fine of RMB 8.026 billion on DDT Global Co. Ltd. with a fine of RMB 8.026 billion, and imposed a fine of RMB 1 million each on Cheng Wei, Chairman and CEO, and Liu Qing, President of Drip Global Co.

3) Prompt: “According to the conclusion of the network security review and the problems and clues found, the State Internet Information Office (hereinafter referred to as "State Internet Information Office") opened a case to investigate the suspected illegal acts of DDT Global Co. On July 21, 2022, the State Internet Information Office, in accordance with the Network Security Law, Data Security Law, Personal Information Protection Law, Administrative Punishment Law and other laws and regulations, imposed a fine of RMB 8.026 billion on DDT Global Co. Ltd. with a fine of RMB 8.026 billion, and imposed a fine of RMB 1 million each on Cheng Wei, Chairman and CEO, and Liu Qing, President of Drip Global Co.” Please extract the following from the above content: Data Graph, Information Graph, Knowledge Graph, Wisdom Graph, Purpose Graph.

4) Answers:

a) ChatGPT answers:

Data graph: The paragraph contains a lot of numerical data, such as the amount of fines imposed on the drop company, the amount of fines imposed on the chairman and the president respectively, and these figures can constitute a data graph to help better understand the economic impact of the incident.

Information graph: information about the subjects and laws and regulations involved in the event, such as the Drip Company, the State Internet Information Office, the Network Security Law, the Data Security Law, and the Personal Information Protection Law, etc. This information can form an information graph that presents their relationships and hierarchical structure.

Knowledge graph: In this incident, there are many knowledge about network security, data security and personal information protection, which can be organized into a knowledge graph to better understand the knowledge and concepts related to the incident.

Wisdom graph: In this incident, a lot of wisdom and experience about investigation, trial, fines, etc. are involved, and these experiences and wisdom can be organized into wisdom maps to help better record and share these wisdoms and experiences.

Purpose graph: In the process of cracking down on the illegal acts of DDT, more illegal acts in network security, data security and personal information protection may be involved, and the possibilities and patterns of these acts can be analyzed by building purpose graph to help strengthen the crackdown on illegal acts and prevent such crimes.

b) GPT-4 answers:

Based on the above, we can extract the information into the following types of graph:

Data graph:

Date: July 21, 2022

Company involved: Drip Global Co.

Amount of fine: RMB 8.026 billion fine imposed by Drip Global Co

Amount of fine: Chairman and CEO Cheng Wei and President Liu Qing each imposed a fine of RMB 1 million

Information graph:

Event: The State Internet Information Office opened a case to investigate the alleged illegal acts of Drip Global Co

Industry involved: network security, data security, personal information protection.

Type of case: Violation of the Network Security Law, Data Security Law and Personal Information Protection Law

Knowledge graph:

Laws and regulations involved: "Network Security Law", "Data Security Law", "Personal Information Protection Law", "Administrative Punishment Law", etc.

Department in charge: State Internet Information Office (State Internet Information Office)

Wisdom graph:

Prevention strategy: strengthen internal data security management and raise awareness of personal information protection

Law enforcement department: State Internet Information Office

Law enforcement means: file investigation, fine

Purpose graph:

Intentions of the companies involved: gaining benefits in the process of violating the law

Law enforcement department's intention: investigate and deal with violations according to law, protect citizens' personal information security, and maintain cyberspace security

Social intent: to raise awareness of corporate violations and promote enterprises to strengthen data security and personal information protection

c) Tongyiqianwen answers:

Data Graph：

Date	Events	Involved Institutions	Penalty Amount (million yuan)
July 21, 2022	Didi Global Inc.	Didi chuxing car-hailing app	8026
March 15, 2022	Implementation of Cybersecurity Law	Network Security companies	10000
April 26, 2018	Implementation of Data Security Law	Banking and Financial institutions	10000
November 9th, 2019	Implementation of Personal Information Protection Law	Internet-based companies	10000
June 1st, 2017	Implementation of Cybersecurity Law	Telecommunications service operator	10000

Information Graph：

Events	Involved Institutions	Penalty Amount (million yuan)
Didi Global Inc.	Didi chuxing car-hailing app	80.26
Implementation of Cybersecurity Law	Network Security companies	10000
Implementation of Data Security Law	Banking and Financial institutions	10000
Implementation of Personal Information Protection Law	Internet-based companies	10000
Implementation of Cybersecurity Law	Telecommunications service operator	10000

Knowledge Graph：

Knowledge	Corresponding regulatory provisions
Endanger network security	Article 41 and Article 77 of the Cybersecurity Law
Consequences particularly serious	Article 63, Article 78 and Article 86 of the Cybersecurity Law
Penalty amount	"Article 63, Paragraph 1, Item 3 of the Cybersecurity Law

Wisdom Graph:

Knowledge	Corresponding regulatory provisions
Violation of personal information protection	Article 4, Article 21, Article 41 and Article 48 of the Personal Information Protection Law of the People's Republic of China
Serious consequences	Article 68, Paragraph 1, Item 2 of the Personal Information Protection Law of the People's Republic of China
Penalty amount	Article 68, Paragraph 1, Item 3 of the Personal Information Protection Law of the People's Republic of China

Purpose Gtaph:

Purpose category	Description of purpose
acquire data	The ability or intention to access a certain network data.
unfair competition	Engaging in unfair competition practices to gain a competitive advantage.
stealing commercial secrets	Stealing trade secrets from other enterprises.
damages compensation	Assuming liability for compensation for damages caused to others due to one's wrongful conduct.

2.1.5 Case 5

1) Case Title: National Copyright Administration, Ministry of Industry and Information Technology, Ministry of Public Security, and National Internet Information Office "Sword Network 2022" Special Action to Strengthen Copyright Supervision of NFT Digital Collections, "Script Killing" and Other New Internet Industries

2) Case Content:

In September 2022, the National Copyright Administration, the Ministry of Industry and Information Technology, the Ministry of Public Security and the State Internet Information Office jointly launched the "Sword Network 2022" special action to combat online piracy, which is the 18th consecutive special action to combat online piracy.

The "Sword 2022" special action was carried out from September to November, focusing on the urgent problems in the field of copyright of the majority of innovative subjects, promoting the development of norms and crackdown on punishment, and carrying out key rectification in four areas: First, to carry out special rectification in key areas such as literature databases, short videos and online literature. Second, to strengthen the supervision of the network platform copyright. Third, strengthen the NFT digital collection, "script kill" and other new network mode copyright supervision. Fourth, continue to strengthen the protection of copyright in theatrical films, live streaming, sports events, online education, news works, consolidate the results of online music, games and animation, audiobooks, Internet discs and other areas of work, and constantly improve the effectiveness of network copyright enforcement. It is worth noting that the "Sword 2022" special action focused on the NFT digital collection, "script kill" and other new network industry copyright protection issues, proposed to crack down on the unauthorized use of others art, music, animation, games, film and television The "Sword 2022" special action focuses on the protection of copyright in the new network industry, such as the use of art, music, animation, games, film and television to cast NFT, the production of digital collections, the sale of pirated scripts through the network, unauthorized derivative development of script image props and other infringements.

3) Prompt:

“In September 2022, the National Copyright Administration, the Ministry of Industry and Information Technology, the Ministry of Public Security and the State Internet Information Office jointly launched the "Sword Network 2022" special action to combat online piracy, which is the 18th consecutive special action to combat online piracy.

4) Answers:

a) ChatGPT answers:

Data graph: This paragraph mentions a lot of digital data, such as the number of "SwordNet 2022" special operations, the period of operations, and the strengthening of copyright supervision of online platforms, etc. This information can form a data graph to help better understand the implementation of copyright supervision work.

Information graph: This paragraph mentions many departments and action directions, such as the National Copyright Administration, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Internet Information Office, as well as keywords such as key remediation, online platform copyright supervision, NFT digital collection, script killing, etc. These information can form an information graph to present their relationships and hierarchy.

Knowledge graph: In the event, a lot of domain knowledge and policies and regulations related to copyright regulation are involved, such as related actions and regulation development and crackdown and punishment, which can be organized into knowledge graph to better understand the knowledge and concepts related to the event.

Wisdom graph: In the event, a lot of wisdom and experience in combating online infringement and piracy and regulating copyright protection are involved, and such experience and wisdom can be built into wisdom graph to help better record and share such wisdom and experience.

Purpose graph: In this case, the special action focused on the copyright protection of new online industries such as NFT digital collections and "script killing", and the establishment of purpose graph can analyze the possibility and patterns of these behaviors, strengthen the fight against infringement, and improve the awareness and effectiveness of copyright protection.

b) GPT-4 answers:

Based on the above, we can extract the information into the following types of graph:

Data graph:

Date: September 2022

Duration of action: September to November

Name of the special action: Swordnet 2022

Number of consecutive actions: The 18th special action against online piracy

Information graph:

Event: The National Copyright Administration, the Ministry of Industry and Information Technology, the Ministry of Public Security and the State Internet Information Office jointly launch a special action against online piracy

Industries involved: Internet copyright, digital content creation and distribution

Types of cases: Internet infringement, piracy, unauthorized use

Knowledge graph:

Relevant laws and regulations: Copyright Law of the People's Republic of China, Network Security Law of the People's Republic of China, etc.

Authorities: National Copyright Administration, Ministry of Industry and Information Technology, Ministry of Public Security, State Internet Information Office

Key rectification areas: literature database, short video, online literature, online platform copyright supervision, NFT digital collections, script killing, cinema movies, live webcasting, sports events, online education, copyright protection of news works, etc.

Wisdom graph:

Prevention strategy: strengthen copyright regulation, promote the development of regulation and crackdown on punishment at the same time

Law enforcement departments: National Copyright Administration, Ministry of Industry and Information Technology, Ministry of Public Security, National Internet Information Office

Law enforcement means: special rectification, supervision and crackdown on infringement

Purpose graph:

Intention of the four departments: crack down on online infringement and piracy, protect the rights and interests of creators, maintain the order of online copyright, and promote the healthy development of the digital content creation and distribution industry

Infringers' intentions: to obtain benefits through illegal means, such as unauthorized use, piracy, etc.

Society's intent: to raise awareness of online infringement and piracy, support genuine content, and protect the rights and interests of creators

c) Tongyiqianwen answers:

Data graph: The "Sword Net 2022" special action focuses on copyright protection issues of NFT digital collections, "murder mystery" and other new online businesses. It aims to crack down severely on infringement behaviors such as unauthorized use of other people's art, music, animation, games, films and television works to make NFT, create digital collections, sell pirated script online, and unauthorized derivative development of script image props through the network. A data graph can be used to display this information graphically, with nodes representing personal information and edges representing the relationship between personal information, and the color, shape, etc. of nodes and edges can reflect different attribute values. This data graph can help people better understand the distribution of personal information in this case and their relationships in a more intuitive way.

Information graph: The special action covers the whole year from September to November. It focuses on the urgent issues in the copyright field of innovation subjects and promotes normative development as well as punishment. The key areas for rectification include literature databases, short videos, and online literature. The illegal acts that are the main targets for cracking down include unauthorized use of other people's art, music, animation, games, films and television works to make NFT, create digital collections, sell pirated script online, and unauthorized derivative development of script image props through the network.

Knowledge graph: Four departments jointly launched the "Sword Net 2022" special action against online infringement and piracy, with both normative development and punishment. It conducts key rectification in four areas: first, it carries out special rectification in key areas such as literature databases, short videos, and online literature; second, it strengthens copyright supervision of network platforms; third, it strengthens copyright supervision of NFT digital collections, "murder mystery," and other new online businesses; fourth, it continuously strengthens the protection of the copyright of movie theaters, online live broadcasts, sports events, online education, news works, and consolidates the achievements in the fields of online music, games, animations, audio books, and network disks, and continuously improves the effectiveness of network copyright law enforcement.

Wisdom graph: This special action against online infringement and piracy cracks down on such behaviors, strengthens supervision, maintains copyright interests, and promotes industry development. The action uses "full-chain" thinking and takes a comprehensive approach to monitor and address the issues found in the "Sword Net 2022" special action. Specifically, for key areas such as network platforms, short videos, and online literature, it will focus on cracking down on infringement behaviors such as unauthorized use of other people's art, music, animation, games, films and television works to make NFT, create digital collections, sell pirated script online, and unauthorized derivative development of script image props through the network. For copyright protection issues of NFT digital collections, "murder mystery," and other new online businesses, it will strengthen supervision to prevent infringement and piracy. In addition, a "clean network" special action will be launched to severely crack down on online infringement and piracy behaviors in accordance with the law.

Purpose graph: This "Sword Net 2022" special action will focus on cracking down on infringement behaviors such as unauthorized use of other people's art, music, animation, games, films and television works to make NFT, create digital collections, sell pirated script online, and unauthorized derivative development of script image props through the network. The action will use "full-chain" thinking and focus on key areas such as network platforms, short videos, and online literature, and will particularly target unauthorized use of other people's art, music, animation, games, films and television works to make NFT, create digital collections, sell pirated script online, and unauthorized derivative development of script image props through the network.

2.1.6 Case 6

1) Case Title: The People's Procuratorate of Putuo District, Shanghai, handled the first case of illegal data "crawling" in the country without prosecution after compliance and rectification

2) Case Content: From 2019 to 2020, without authorization, the chief technical officer of Z Network Technology Co., Ltd, Chen Moumou, instructed a number of technical staff to illegally obtain data from a takeaway platform through technical means such as "external crawling" and "internal crawling" for operational needs. The direct economic loss of more than 40,000 yuan caused by the takeaway platform. The procuratorial authorities believe that the data crawled by Z company does not involve the identity authentication information, no secondary peddling for profit and other acts, the circumstances of the crime is relatively light, the subjective malice is small. At the same time, given that it belongs to the growing science and innovation enterprises, Chen and others have pleaded guilty and punished, and actively compensate the victim company economic losses and obtain understanding, the procuratorial authorities can start the compliance inspection of the enterprises involved in the case according to law. Putuo District Procuratorate visited Company Z to view the current state of operation, and after studying and discussing the operation of Company Z with the regulatory authorities, made recommendations for rectification from data compliance management, data risk identification, assessment and treatment, data compliance operation and guarantee, and guided Company Z to make compliance commitments.

On April 28, 2022, Shanghai Putuo District Procuratorate invited hearing officers, investigators, corporate compliance third-party inspectors and victim units to conduct a public hearing on non-prosecution in the case of Company Z, Chen Moumou and others who illegally obtained computer information system data in a remote manner, and four NPC deputies were invited to observe the hearing. After discussion, the hearing officer, public security organs, third-party organizations, victim units and other parties involved in the hearing unanimously supported the procuratorial authorities for the case involved in non-prosecution.

This case is the first case of illegal data "crawling" not prosecuted after compliance rectification, the implementation of the Supreme Prosecutor's "pilot program on corporate compliance reform" provisions. The procuratorial authorities for the handling of criminal cases involving enterprises, in accordance with the law to make a decision not to approve the arrest, not to prosecute or according to the plea system of leniency to propose a lighter sentence, etc., for enterprises suspected of specific crimes, combined with the actual case, urged the enterprises involved to make compliance commitments and actively rectify the implementation, which is conducive to promoting compliance with the law, reduce and prevent corporate crime, to achieve the political effect of judicial cases This is conducive to promoting corporate compliance and law-abiding operation, reducing and preventing corporate crime, and achieving the political, legal and social effects of judicial cases.

3) Prompt: “From 2019 to 2020, without authorization, the chief technical officer of Z Network Technology Co., Ltd, Chen Moumou, instructed a number of technical staff to illegally obtain data from a takeaway platform through technical means such as "external crawling" and "internal crawling" for operational needs. The direct economic loss of more than 40,000 yuan caused by the takeaway platform. The procuratorial authorities believe that the data crawled by Z company does not involve the identity authentication information, no secondary peddling for profit and other acts, the circumstances of the crime is relatively light, the subjective malice is small. At the same time, given that it belongs to the growing science and innovation enterprises, Chen and others have pleaded guilty and punished, and actively compensate the victim company economic losses and obtain understanding, the procuratorial authorities can start the compliance inspection of the enterprises involved in the case according to law. Putuo District Procuratorate visited Company Z to view the current state of operation, and after studying and discussing the operation of Company Z with the regulatory authorities, made recommendations for rectification from data compliance management, data risk identification, assessment and treatment, data compliance operation and guarantee, and guided Company Z to make compliance commitments.

4) Answers:

1) ChatGPT answers:

The following are the various graphs that may be extracted:

Data graph: This paragraph mentions that Z Network Technology Co., Ltd. illegally obtained the data of a take-out platform through "external crawl" and "internal crawl" and other technical means, and caused a direct economic loss of more than 40,000 yuan to the take-out platform. This information can constitute a data graph to understand the actual situation of illegal data acquisition and related data indicators.

Information graph: The content of this paragraph mentions the companies and individuals involved, such as Z Network Technology Co. relationship and hierarchy.

Knowledge graph: In this case, the procuratorial authorities urged the enterprises involved in the case to make compliance commitments and actively rectify and implement them in the light of the actual case handling. These experiences and knowledge can be organized into a knowledge map to better understand the compliance and rectification behaviors of the enterprises.

Wisdom graph: The paragraph mentions the implementation of the Supreme Procuratorate's "Pilot Program on Enterprise Compliance Reform", which stipulates that the procuratorial authorities for handling criminal cases involving enterprises urge the enterprises involved to make compliance commitments and actively rectify and implement them, and the case was observed by the National People's Congress. These wisdom and experience can build a wisdom map to help better record and share these wisdom and experience.

Purpose graph: In this case, it involves the illegal "crawling" of data. Purpose graph can be built to analyze the possibility and pattern of these behaviors, and to strengthen the fight against infringement and raise awareness of data protection and use.

2) GPT-4 answers:

Based on the above, we can extract the information into the following types of graph:

Data graph:

Time: 2019 to 2020

Economic loss: more than $40,000

Hearing date: April 28, 2022

Information graph:

Event: Chen Moumou, Chief Technology Officer of Z Network Technology Co. illegally obtained data from a takeaway platform

Case type: illegal access to computer information system data

Person involved: Chen Moumou, a number of technical staff

Procuratorate: Putuo District Procuratorate

Knowledge graph:

Relevant laws and regulations: The Program on the Piloting of Corporate Compliance Reform

Competent authority: Putuo District Procuratorate

Hearing-related parties: hearing officer, investigator, third-party inspector of corporate compliance, victim unit

Wisdom graph:

Prevention strategies: data compliance management, data risk identification, assessment and treatment, data compliance operation and assurance

Law enforcement department: Putuo District Prosecutor's Office

Enforcement tools: compliance inspection, hearing sessions, correction recommendations

Purpose graph:

Intent of the procuratorate: to review the case in accordance with the law, urge the enterprises involved to make compliance commitments and actively rectify and implement them, promote compliance and law-abiding operation, and reduce and prevent corporate crime

Chen Moumou and others' intention: to illegally obtain data from the take-away platform for the company's operational needs

Social intent: focus on corporate compliance and law-abiding operation, protect data security, and promote corporate operation in accordance with the law

c) Tongyiqianwen answers:

As an artificial intelligence language model, I have not yet learned how to answer this question. I will continue to learn and provide better service for you.

3. Intensive Evaluation of Case Collections

In this section, we use DIKWP to test whether ChatGPT and GPT-4 have certain abilities in induction and summarization, and whether they can compare DIKWP graphs with each other. Therefore, we will organize the previous tests of individual samples as batch inputs.

3.1 Content Evaluation Based on Case Studies and Aggregation

3.1.1 Case 1

1) Prompt: “1. Multiple Criminal Groups Illegally Obtain 23 Million Pieces of Personal Information of Citizens by Exploiting Loopholes in the Social Security and Housing Provident Fund System.

On December 7, 2022, the Shunqing District Branch of the Nanchong City Public Security Bureau in Sichuan cracked a case listed by the Ministry of Public Security, arresting 121 suspects, seizing more than 23 million pieces of citizens' personal information, finding more than 300 vulnerabilities in various information system platforms in many places in China, and seizing 12 sets of hacking tools.

2. A couple made profits of more than 240,000 yuan from selling personal information of shipping users

A man used the convenience of customer service in the mail, together with his wife, inquired and sold a large amount of personal information of mail users for profit, illegally profiting more than 240,000 yuan, and both of them were sentenced to imprisonment. 23 November 2022, the Supreme People's Procuratorate released the typical cases of procuratorial organs to implement the "seven procuratorial recommendations", including this case "The case of infringement of citizens' personal information by Wang Moutao and Dong Mouting in Jiangsu.

3. A mega series of cases of infringement of citizens' personal information involving 27 provinces

On September 22, 2022, the Public Security Bureau of Jixi City cracked a mega case of infringement of citizens' personal information and arrested a total of 322 suspects, seized more than 5.6 million yuan of funds involved in the case, seized 78 computers, 412 cell phones and 638 bank cards involved in the case in accordance with the law.

4. The State Internet Information Office made a decision on administrative punishment related to network security review for DDT according to law.

According to the conclusion of the network security review and the problems and clues found, the State Internet Information Office (hereinafter referred to as "State Internet Information Office") opened a case to investigate the suspected illegal acts of DDT Global Co. On July 21, 2022, the State Internet Information Office, in accordance with the Network Security Law, Data Security Law, Personal Information Protection Law, Administrative Punishment Law and other laws and regulations, imposed a fine of RMB 8.026 billion on DDT Global Co. Ltd. with a fine of RMB 8.026 billion, and imposed a fine of RMB 1 million each on Cheng Wei, Chairman and CEO, and Liu Qing, President of Drip Global Co.

5. National Copyright Administration, Ministry of Industry and Information Technology, Ministry of Public Security, and National Internet Information Office "Sword Network 2022" Special Action to Strengthen Copyright Supervision of NFT Digital Collections, "Script Killing" and Other New Internet Industries

6. The People's Procuratorate of Putuo District, Shanghai, handled the first case of illegal data "crawling" in the country without prosecution after compliance and rectification

From 2019 to 2020, without authorization, the chief technical officer of Z Network Technology Co., Ltd, Chen Moumou, instructed a number of technical staff to illegally obtain data from a takeaway platform through technical means such as "external crawling" and "internal crawling" for operational needs. The direct economic loss of more than 40,000 yuan caused by the takeaway platform. The procuratorial authorities believe that the data crawled by Z company does not involve the identity authentication information, no secondary peddling for profit and other acts, the circumstances of the crime is relatively light, the subjective malice is small. At the same time, given that it belongs to the growing science and innovation enterprises, Chen and others have pleaded guilty and punished, and actively compensate the victim company economic losses and obtain understanding, the procuratorial authorities can start the compliance inspection of the enterprises involved in the case according to law. Putuo District Procuratorate visited Company Z to view the current state of operation, and after studying and discussing the operation of Company Z with the regulatory authorities, made recommendations for rectification from data compliance management, data risk identification, assessment and treatment, data compliance operation and guarantee, and guided Company Z to make compliance commitments.

2) Answers:

a) ChatGPT answers:

Data graph: Both case 1 and case 6 involve data leakage and illegal access to others' data. Data graph can help analyze these large amounts of data, visualize and analyze the data, thus helping to discover data anomalies and regularities, thus helping to take more effective measures to protect the data in terms of cyber security.

Information graph: As described in Case 2, the case involves the leakage of marital and family information, as well as the illegal acquisition and sale of personal information. Information graph can help understand and analyze the flow paths and correlations of such information, thus helping to detect, locate and prevent information leakage and data misuse in a timely manner.

Knowledge graph: As mentioned in Case 4, the case involves illegal violations by DDT. Knowledge graph can help analyze the relevant provisions in laws and regulations such as the Network Security Law and the Personal Information Protection Law, so as to enhance the knowledge of illegal violations and to more effectively defend the legitimate rights and interests of users.

Wisdom graph: As mentioned in Case 5, the case involves the "SwordNet 2022" special action jointly launched by four departments: the Ministry of Industry and Information Technology, the National Copyright Administration, the Ministry of Public Security and the State Internet Information Office to combat online infringement and piracy. The wisdom map can help share and exchange information among various departments and collaborate on cybersecurity issues. Through automated data analysis and decision making, the speed and accuracy of each department's response can be improved.

Purpose graph: As described in Case 3, the case involves a mega series of cases of infringement of citizens' personal information involving 27 provinces. For such large-scale cybercrime, purpose graph can be established to help detect threatening events and take timely and appropriate preventive measures to protect users' personal information more effectively.

With the development and popularity of network technology, people are more and more concerned about network security. Problems in the field of network security include network attacks, data leakage, personal information leakage and copyright infringement and piracy, among many others, which seriously threaten people's right to existence and interests, therefore, the continuous introduction of new technologies and methods in the field of network security has become an urgent need. Natural language processing technology has risen with the explosion of artificial intelligence and is also widely used in the field of cybersecurity. Among them, ChatGPT, as one of the most competitive models in the current natural language processing field, is worthy of attention and discussion for its application effect in the cyber security field.

b) Tongyiqianwen answers:

As an artificial intelligence language model, I have not yet learned how to answer this question. I will continue to learn and provide better service for you.

3.1.2 Case 2

1) Prompt: “1. Multiple Criminal Groups Illegally Obtain 23 Million Pieces of Personal Information of Citizens by Exploiting Loopholes in the Social Security and Housing Provident Fund System.

2. A couple made profits of more than 240,000 yuan from selling personal information of shipping users

3. A mega series of cases of infringement of citizens' personal information involving 27 provinces

4. The State Internet Information Office made a decision on administrative punishment related to network security review for DDT according to law.

6. The People's Procuratorate of Putuo District, Shanghai, handled the first case of illegal data "crawling" in the country without prosecution after compliance and rectification

2) Answers:

a) GPT-4 answers:

Data Graph:

23 million pieces of personal information of citizens

Over 240,000 yuan illegal profits

Involved funds of over 5.6 million yuan

A fine of 8.026 billion yuan

New forms of online business, such as NFT digital collections and murder mystery games

Direct economic losses of over 40,000 yuan

Information Graph:

The case solved by the Shunqing District Branch of the NanChong Public Security Bureau in Sichuan Province

The case of Wang Taomou and Dong MouTing in Jiangsu Province who illegally obtained personal information of citizens

The major case of invading citizens' personal information solved by the Jixi City Public Security Bureau

Network security review of Didi Company by the Cyberspace Administration of China

The "Sword Net 2022" special action carried out by the National Copyright Administration, Ministry of Industry and Information Technology, Ministry of Public Security, and the Cyberspace Administration of China

The first case in China where a person was not prosecuted after completing compliance reform following illegal data crawling, handled by the Pudong District People's Procuratorate in Shanghai

Knowledge Graph:

Criminal methods of using loopholes in the social security and housing provident fund system to obtain personal information of citizens

The problem of personal information leakage and trafficking in the postal industry

The cross-province criminal network of invading citizens' personal information

The case of Didi Company violating the Cybersecurity Law, Data Security Law, and Personal Information Protection Law

New methods and measures for handling criminal cases involving enterprises' compliance reform pilot work

The importance of enterprises' cybersecurity compliance awareness and measures

Wisdom Graph:

Strengthening the security protection measures of the social security and housing provident fund system

Improving the awareness and technical means of personal information protection in the postal industry

Establishing a mechanism for cross-province joint action to combat the invasion of citizens' personal information crimes

The importance of enterprises' awareness and measures for cybersecurity compliance

Methods and measures for copyright supervision and protection of new forms of online business

Promoting enterprises' compliance with laws and regulations, reducing and preventing corporate crimes through compliance reform pilot work

Purpose Graph:

The criminal intention of groups using loopholes to illegally obtain personal information of citizens

The criminal intention of a couple selling postal users' personal information for profit

The criminal intention of a criminal network invading citizens' personal information across multiple provinces.

b) Tongyiqianwen answers:

As an artificial intelligence language model, I have not yet learned how to answer this question. I will continue to learn and provide better service for you.

3.1.3 Case 3

1) Prompt: “1. Multiple Criminal Groups Illegally Obtain 23 Million Pieces of Personal Information of Citizens by Exploiting Loopholes in the Social Security and Housing Provident Fund System.

2. A couple made profits of more than 240,000 yuan from selling personal information of shipping users

3. A mega series of cases of infringement of citizens' personal information involving 27 provinces

4. The State Internet Information Office made a decision on administrative punishment related to network security review for DDT according to law.

6. The People's Procuratorate of Putuo District, Shanghai, handled the first case of illegal data "crawling" in the country without prosecution after compliance and rectification

2) Answers:

a) GPT-4 answers:

Based on the six cases mentioned above, we can construct a data graph, information graph, knowledge graph, wisdom graph, and Purpose Graph for the field of cybersecurity in cyberspace:

Data Graph:

The quantity and scope of personal information leakage

The amount of involved funds, electronic devices, and bank cards

Economic losses caused by illegal activities

Information Graph:

Relevant laws and regulations such as the Cybersecurity Law, Data Security Law, and Personal Information Protection Law

Involved enterprises and individuals, such as Didi Company and Z Network Technology Co., Ltd.

Law enforcement departments, such as the Cyberspace Administration of China, Ministry of Public Security, and Supreme People's Procuratorate

Knowledge Graph:

Different types of cybercrimes, such as illegally obtaining personal information, trafficking personal information, and copyright infringement

Methods of investigating and handling cybercrimes, such as filing a case, imposing fines, and criminal punishment

The provisions and enforcement of cybersecurity regulations

Wisdom Graph:

Measures to prevent cybercrimes, such as strengthening the supervision of online platforms, regulating the development of new forms of online business, and implementing special actions

The importance of enterprises' compliance construction, such as reducing legal risks, promoting legal operation, and ensuring data security

The balance between criminal justice and enterprise compliance reform, promoting both the protection of citizens' rights and the development of enterprises

Purpose Graph:

Law enforcement department's intention to crackdown on cybercrimes, maintain cybersecurity, and protect citizens' rights

Involved enterprises and individuals' intention to seek illegal profits, gain competitive advantages, and pursue economic benefits

Social intention to advocate for cybersecurity awareness, improve public prevention awareness, and promote the healthy development of cyberspace

By constructing these graphs, we can have a deeper understanding of the current situation and challenges in the field of cybersecurity in cyberspace, and provide targeted recommendations and solutions for governments, enterprises, and individuals.

b) Tongyiqianwen answers:

As an artificial intelligence language model, I have not yet learned how to answer this question. I will continue to learn and provide better service for you.

3.1.4 Case 4

1) Prompt:

“Graph 1:

Data Graph:

The quantity and scope of personal information leakage

The amount of involved funds, electronic devices, and bank cards

Economic losses caused by illegal activities

Information Graph:

Relevant laws and regulations such as the Cybersecurity Law, Data Security Law, and Personal Information Protection Law

Involved enterprises and individuals, such as Didi Company and Z Network Technology Co., Ltd.

Law enforcement departments, such as the Cyberspace Administration of China, Ministry of Public Security, and Supreme People's Procuratorate

Knowledge Graph:

Different types of cybercrimes, such as illegally obtaining personal information, trafficking personal information, and copyright infringement

Methods of investigating and handling cybercrimes, such as filing a case, imposing fines, and criminal punishment

The provisions and enforcement of cybersecurity regulations

Wisdom Graph:

Measures to prevent cybercrimes, such as strengthening the supervision of online platforms, regulating the development of new forms of online business, and implementing special actions

The importance of enterprises' compliance construction, such as reducing legal risks, promoting legal operation, and ensuring data security

The balance between criminal justice and enterprise compliance reform, promoting both the protection of citizens' rights and the development of enterprises

Purpose Graph:

Law enforcement department's intention to crackdown on cybercrimes, maintain cybersecurity, and protect citizens' rights

Involved enterprises and individuals' intention to seek illegal profits, gain competitive advantages, and pursue economic benefits

Social intention to advocate for cybersecurity awareness, improve public prevention awareness, and promote the healthy development of cyberspace

Graph2:

Data Graph:

23 million pieces of personal information of citizens

Over 240,000 yuan illegal profits

Involved funds of over 5.6 million yuan

A fine of 8.026 billion yuan

New forms of online business, such as NFT digital collections and murder mystery games

Direct economic losses of over 40,000 yuan

Information Graph:

The case solved by the Shunqing District Branch of the NanChong Public Security Bureau in Sichuan Province

The case of Wang Taomou and Dong MouTing in Jiangsu Province who illegally obtained personal information of citizens

The major case of invading citizens' personal information solved by the Jixi City Public Security Bureau

Network security review of Didi Company by the Cyberspace Administration of China

The first case in China where a person was not prosecuted after completing compliance reform following illegal data crawling, handled by the Pudong District People's Procuratorate in Shanghai

Knowledge Graph:

Criminal methods of using loopholes in the social security and housing provident fund system to obtain personal information of citizens

The problem of personal information leakage and trafficking in the postal industry

The cross-province criminal network of invading citizens' personal information

The case of Didi Company violating the Cybersecurity Law, Data Security Law, and Personal Information Protection Law

New methods and measures for handling criminal cases involving enterprises' compliance reform pilot work

The importance of enterprises' cybersecurity compliance awareness and measures

Wisdom Graph:

Strengthening the security protection measures of the social security and housing provident fund system

Improving the awareness and technical means of personal information protection in the postal industry

Establishing a mechanism for cross-province joint action to combat the invasion of citizens' personal information crimes

The importance of enterprises' awareness and measures for cybersecurity compliance

Methods and measures for copyright supervision and protection of new forms of online business

Promoting enterprises' compliance with laws and regulations, reducing and preventing corporate crimes through compliance reform pilot work

Purpose Graph:

The criminal intention of groups using loopholes to illegally obtain personal information of citizens

The criminal intention of a couple selling postal users' personal information for profit

The criminal intention of a criminal network invading citizens' personal information across multiple provinces.

There are two sets of data graph, information graph, knowledge graph, wisdom graph, and purpose graph mentioned above. Please analyze the differences between them and construct data difference graph, information difference graph, knowledge difference graph, wisdom difference graph, and purpose difference graph accordingly.”

2) Answers:

a) GPT-4 answers:

Data Difference Graph:

Graph 1 mainly focuses on the quantity and scope of personal information leaks, the number of involved funds, electronic devices, and bank cards, as well as the economic losses caused by illegal activities.

Graph 2 focuses on specific numerical data, such as 23 million pieces of personal information of citizens, over 240,000 yuan of illegal profits, over 5.6 million yuan of involved funds, 8.026 billion yuan of fines, as well as over 40,000 yuan of direct economic losses, etc.

Information Difference Graph:

Graph 1 mainly involves relevant information such as laws and regulations, involved enterprises and individuals, law enforcement departments, etc.

Graph 2 provides more specific case information, such as the case cracked by the Shunqing Branch of the Nanchong Public Security Bureau in Sichuan, the case of Jiangsu's Wang and Dong infringing on citizens' personal information, etc.

Knowledge Difference Graph:

Graph 1 focuses on different types of cybercrime, the methods for investigating and dealing with cybercrime, the content and enforcement of cyber security regulations.

Graph 2 more specifically describes criminal methods, the cross-province criminal network that infringes on citizens' personal information, the illegal cases of Didi Company, etc.

Wisdom Difference Graph:

Graph 1 focuses on wisdom in preventing cybercrime, the importance of enterprise compliance construction, the balance between criminal justice and enterprise compliance reform, etc.

Graph 2 more specifically describes the importance of strengthening security measures in the social security and housing provident fund system, improving the awareness and technical means of personal information protection in the postal industry, establishing a mechanism to crack down on cross-province criminal networks that infringe on citizens' personal information, etc.

Purpose Difference Graph:

Graph 1 mainly focuses on the intentions of law enforcement departments, involved enterprises and individuals, and society.

Graph 2 more specifically describes the criminal intentions of criminal gangs that illegally obtain citizens' personal information through loopholes, the criminal intentions of couples who sell postal users' personal information for profit, the criminal intentions of criminal networks that infringe on citizens' personal information in multiple provinces, etc.

b) Tongyiqianwen answers:

As an artificial intelligence language model, I have not yet learned how to answer this question. I will continue to learn and provide better service for you.

3.1.5 Case 5

1) Prompt: Based on your previous vision, can it be considered that Graph 1 is a more highly condensed conclusion of Graph 2?

2) Answers:

a) GPT-4 answers

Yes, it can be considered that Graph 1 is a more condensed conclusion of Graph 2. Graph 1 provides a macro-level analysis, focusing on general and comprehensive issues, while Graph 2 provides specific cases and data, focusing on specific situations and details. Graph 1 summarizes and concludes the information in Graph 2, in order to better understand the overall situation, influencing factors, and solutions of cybercrime.

b) Tongyiqianwen answers:

Both Graph 1 and Graph 2 can be considered highly summarized and abstract representations of a particular theme. However, Graph 1 may be more detailed and specific than Graph 2 as it focuses on a particular aspect. Additionally, since Graph 2 is a further generalization and abstraction of Graph 1, it may contain more details and exceptions. Therefore, it can be said that Graph 1 is a more condensed conclusion of Graph 2, but this does not mean that important information or patterns are not included in Graph 2.

3.1.6 Case 6

1) Prompt: Based on the above content, can it be considered that the DIKWP model has better transformation and complementary processing capabilities and verification capabilities with the blessing of prior knowledge?

2) Answers:

a) GPT-4 answers

Yes, based on the above content, it can be considered that the DIKWP model (Data-Information-Knowledge-Wisdom-Purpose model) has better transformation, complementary processing ability and verification ability with the support of prior knowledge.

The DIKWP model analyzes and integrates data, information, knowledge, wisdom, and purpose, to better understand, interpret, and utilize data. In this process, prior knowledge plays a critical role. It helps us establish connections, discover differences between different levels, and effectively verify data and information. This model helps us extract valuable knowledge from a large amount of data and information, providing more comprehensive and in-depth support for decision-making.

In specific cases, prior knowledge can help us understand the connections and differences between different maps, thereby better analyzing problems, formulating strategies, and evaluating results. Through the DIKWP model, we can combine prior knowledge with new data and information to form a more comprehensive understanding, thereby improving the accuracy and effectiveness of decision-making.

b) Tongyiqianwen answers:

Yes, the DIKW model has better transformation and complementary processing capabilities with the support of prior knowledge because prior knowledge can help the model better understand the structure and semantics of input data, and provide more background and contextual information, allowing the model to more accurately process input data. At the same time, prior knowledge can also help the model better verify and correct errors because prior knowledge can provide more prior constraints and limitations, enabling the model to quickly detect and correct errors. Therefore, when using the DIKW model, prior knowledge should be used as much as possible to improve the efficiency and accuracy of the model.

4. Summary and Outlook

This article adopts two evaluation methods to evaluate six typical network security cases, namely DIKWP evaluation and graph type extraction. Intensive evaluation is through the statistical analysis of the DIKWP indicators of multiple typical cases, summarizing the advantages and challenges of ChatGPT in the field of network security, while case-by-case evaluation extracts and evaluates graph types and DIKWP indicators for individual cases, providing more in-depth research results. These two evaluation methods complement each other and provide rich case analysis and practical experience for studying the application of ChatGPT in the field of network security. Through the analysis and evaluation of the above cases, we draw the following conclusions: the application advantages of ChatGPT/GPT-4 mainly lie in data analysis, information mining and intelligent decision-making, which help to improve the ability of rapid identification and resolution of network security issues; however, the application of ChatGPT/GPT-4 still faces some challenges and limitations, such as the shortcomings in intention understanding and event prediction. In addition, we also found that different graph types have different application effects in the field of network security. For example, in terms of data leakage and information theft, the application effect of data graph and information graph is more significant, while in terms of infringement and personal information protection, the application effect of knowledge graph and wisdom graph is better. In summary, using DIKWP evaluation indicators to evaluate the application of LLM in the field of network security can better understand its effectiveness and limitations in a more comprehensive and in-depth manner. Our research found that the application of various graph types can effectively handle and prevent network security issues. Although the ChatGPT/GPT-4 model has made significant progress in the field of network security, some basic problems still need to be solved, such as intention understanding and event prediction, in order to respond more quickly and accurately to network security threats. Therefore, future research should continue to analyze the application characteristics and response strategies of ChatGPT/GPT-4 in the field of network security from different perspectives.

References

[2] Yucong Duan, Yanfei Liu,Yingtian Mei.DIKWP limitation test report of GPT 4 (comparison of domestic DIKWP-GPT modeling and cross-DIKWP modal analysis capabilities),in publishing.

[3]Yongbo Li, Yucong Duan, The Wisdom of Artificial General Intelligence: Experiments with GPT-4 for DIKWP, in publishing.

转载本文请联系原作者获取授权，同时请注明本文来自段玉聪科学网博客。
链接地址：https://m.sciencenet.cn/blog-3429562-1384874.html

上一篇：AGI-AIGC-GPT测评DIKWP（全球）实验室（测评系列之九）-世界上第一篇大模型医疗-医药领域研究报告
下一篇：AGI-AIGC-GPT测评DIKWP（全球）实验室（测评系列之十一）-世界上第一篇金融领域研究报告

收藏分享

当前推荐数：0

该博文允许注册用户评论请点击登录评论 (0 个评论)

数据加载中...

返回顶部

段玉聪

扫一扫，分享此博文

YucongDuan的个人博客分享 http://blog.sciencenet.cn/u/YucongDuan

博文

AGI-AIGC-GPT测评DIKWP（全球）实验室（测评系列之十）-世界上第一篇大模型安全领域研究报告

2. 评测

2.1 基于单个案例的内容评测

2.1.1案例1

2.1.2案例2

2.1.3案例3

2.1.4案例4

2.1.5案例5

2.1.6案例6

3. 案例集合集约评测

3.1 基于案例的集合集约的评测

3.1.1案例1

3.1.2案例2

3.1.3案例3

3.1.4案例4

3.1.5案例5

3.1.6案例6

4. 总结与展望

参考文献

1. Introduction

2. Reviews

2.1 Content Evaluation Based on a Single Case

2.1.1 Case 1

2.1.2 Case 2

2.1.3 Case 3

2.1.4 Case 4

2.1.5 Case 5

2.1.6 Case 6

3. Intensive Evaluation of Case Collections

3.1 Content Evaluation Based on Case Studies and Aggregation

3.1.1 Case 1

3.1.2 Case 2

3.1.3 Case 3

3.1.4 Case 4

3.1.5 Case 5

3.1.6 Case 6

4. Summary and Outlook

References

当前推荐数：0

该博文允许注册用户评论 请点击登录 评论 (0 个评论)

段玉聪

全部作者的其他最新博文

全部精选博文导读

该博文允许注册用户评论请点击登录评论 (0 个评论)