In the digital age, the global character of the Internet has significantly improved our daily lives by providing access to large amounts of knowledge and allowing for seamless connections. However, this enormously int...In the digital age, the global character of the Internet has significantly improved our daily lives by providing access to large amounts of knowledge and allowing for seamless connections. However, this enormously interconnected world is not without its risks. Malicious URLs are a powerful menace, masquerading as legitimate links while holding the intent to hack computer systems or steal sensitive personal information. As the sophistication and frequency of cyberattacks increase, identifying bad URLs has emerged as a critical aspect of cybersecurity. This study presents a new approach that enables the average end-user to check URL safety using Microsoft Excel. Using the powerful VirusTotal API for URL inspections, this study creates an Excel add-in that integrates Python and Excel to deliver a seamless, user-friendly interface. Furthermore, the study improves Excel’s capabilities by allowing users to encrypt and decrypt text communications directly in the spreadsheet. Users may easily encrypt their conversations by simply typing a key and the required text into predefined cells, enhancing their personal cybersecurity with a layer of cryptographic secrecy. This strategy democratizes access to advanced cybersecurity solutions, making attentive digital integrity a feature rather than a daunting burden.展开更多
近年来,使用恶意Excel 4.0宏(XLM)文档的攻击迎来了爆发,而XLM代码往往经过复杂的混淆,现有方法或检测系统难以分析海量样本的真实功能。因此,针对恶意样本中使用的各类混淆技术,基于抽象语法树和模拟执行,设计和实现了包含138个宏函数...近年来,使用恶意Excel 4.0宏(XLM)文档的攻击迎来了爆发,而XLM代码往往经过复杂的混淆,现有方法或检测系统难以分析海量样本的真实功能。因此,针对恶意样本中使用的各类混淆技术,基于抽象语法树和模拟执行,设计和实现了包含138个宏函数处理程序的自动化XLM反混淆与关键威胁指标(IOC,indicators of compromise)提取系统XLMRevealer;在此基础上,根据XLM代码特点提取Word和Token特征,通过特征融合能够捕获多层次细粒度特征,并在XLMRevealer中构造CNN-BiLSTM(convolution neural network-bidirectional long short term memory)模型,从不同维度挖掘家族样本的关联性和完成家族分类。最后,从5个来源构建包含2346个样本的数据集并用于反混淆实验和家族分类实验。实验结果表明,XLMRevealer的反混淆成功率达到71.3%,相比XLMMacroDeobfuscator和SYMBEXCEL工具分别提高了20.8%和15.8%;反混淆效率稳定,平均耗时仅为0.512 s。XLMRevealer对去混淆XLM代码的家族分类准确率高达94.88%,效果优于所有基线模型,有效体现Word和Token特征融合的优势。此外,为探索反混淆对家族分类的影响,并考虑不同家族使用的混淆技术可能有所不同,模型会识别到混淆技术的特征,分别对反混淆前和反混淆后再统一混淆的XLM代码进行实验,家族分类准确率为89.58%、53.61%,证明模型能够学习混淆技术特征,更验证了反混淆对家族分类极大的促进作用。展开更多
文摘In the digital age, the global character of the Internet has significantly improved our daily lives by providing access to large amounts of knowledge and allowing for seamless connections. However, this enormously interconnected world is not without its risks. Malicious URLs are a powerful menace, masquerading as legitimate links while holding the intent to hack computer systems or steal sensitive personal information. As the sophistication and frequency of cyberattacks increase, identifying bad URLs has emerged as a critical aspect of cybersecurity. This study presents a new approach that enables the average end-user to check URL safety using Microsoft Excel. Using the powerful VirusTotal API for URL inspections, this study creates an Excel add-in that integrates Python and Excel to deliver a seamless, user-friendly interface. Furthermore, the study improves Excel’s capabilities by allowing users to encrypt and decrypt text communications directly in the spreadsheet. Users may easily encrypt their conversations by simply typing a key and the required text into predefined cells, enhancing their personal cybersecurity with a layer of cryptographic secrecy. This strategy democratizes access to advanced cybersecurity solutions, making attentive digital integrity a feature rather than a daunting burden.
文摘近年来,使用恶意Excel 4.0宏(XLM)文档的攻击迎来了爆发,而XLM代码往往经过复杂的混淆,现有方法或检测系统难以分析海量样本的真实功能。因此,针对恶意样本中使用的各类混淆技术,基于抽象语法树和模拟执行,设计和实现了包含138个宏函数处理程序的自动化XLM反混淆与关键威胁指标(IOC,indicators of compromise)提取系统XLMRevealer;在此基础上,根据XLM代码特点提取Word和Token特征,通过特征融合能够捕获多层次细粒度特征,并在XLMRevealer中构造CNN-BiLSTM(convolution neural network-bidirectional long short term memory)模型,从不同维度挖掘家族样本的关联性和完成家族分类。最后,从5个来源构建包含2346个样本的数据集并用于反混淆实验和家族分类实验。实验结果表明,XLMRevealer的反混淆成功率达到71.3%,相比XLMMacroDeobfuscator和SYMBEXCEL工具分别提高了20.8%和15.8%;反混淆效率稳定,平均耗时仅为0.512 s。XLMRevealer对去混淆XLM代码的家族分类准确率高达94.88%,效果优于所有基线模型,有效体现Word和Token特征融合的优势。此外,为探索反混淆对家族分类的影响,并考虑不同家族使用的混淆技术可能有所不同,模型会识别到混淆技术的特征,分别对反混淆前和反混淆后再统一混淆的XLM代码进行实验,家族分类准确率为89.58%、53.61%,证明模型能够学习混淆技术特征,更验证了反混淆对家族分类极大的促进作用。
