BERT-Enhanced SQL Injection Black-Box Detection (SqliBERT)

The new system is essentially a significant upgrade to
the current systems in a way that it can locate and manage the assaults on SQL Injection by application of language
representation technology. These are advanced systems that do not merely rely on highly specific patterns or sets of rules for the most part. BERT-Enhanced SQL Injection Black-Box Detection (SqliBERT) employs a pre-trained Bidirectional Encoder Representations from Transformers model, which is fine-tuned to obtain the semantic and syntactic structures of the input query. In other words, the system is able to find those tiny modifications and concealed injections that attempt to evade signatures and heuristics-based tools, and also it gives a precision score of 93% thus it keeps the system stability at a high level. The most important point to bring home with this method is that it can automatically acquire the contextual features from the query sequences; therefore, it can very easily distinguish the
normal ones from the harmful ones in a black-box scenario
where there is no information about the internal logic of the system. Besides this, black-box testing is also included here; thus, it is a more realistic approach for the cases in the real world where the application cannot be accessed directly. The outlined model might be the best if it couples contextual embeddings with open learning.
Keywords— BERT, SQL Injection, Black-Box Detection, Deep
Learning, Cybersecurity, Natural Language Processing, Query
Analysis, Database Security, Contextual Embeddings, Intrusion Detection.