Deep Learning Based Software Vulnerability Detection in Code Snippets and Tag Questions Using Convolutional Neural Networks

With the increase in the usage of the internet for gaining and providing information and knowledge, questionnaire forums are becoming popular means of the same for both teenagers and adults. Users post questions and mark them with the topic that they are related to (known as tags). Post this, users always expect quick answers/solutions from known and reliable sources. However, a large number of posted questions remain unanswered due to erroneous and huge number of tags. In our system, we propose an automated method to generate these tags using machine learning and deep learning techniques. This mainly helps in standardizing the tag content for similar questions and the total number of tags that the system needs to deal with. These tags can be generated from the content of the question provided by the user. In addition to this, for questionnaire forums related to programming questions like StackOverflow and StackExchange, our system aims to use the code snippets provided by the user to detect some common software vulnerabilities using deep learning techniques. Software vulnerabilities are weaknesses in software that may be exploited by malicious individuals. These vulnerabilities can arise from coding errors, flaws in design, or inadequate security measures, and can result in unauthorized access, data breaches, or system availability. Detection of these vulnerabilities beforehand will benefit the user and prevent any sort of security threats in their system.