Program Dependency Graphs (PDGs) have been a crucial component of software engineering for many years, enabling numerous activities including program slicing, security analysis, debugging, and optimization of code. Based on this idea, Database-Oriented Program Dependency Graphs (DOPDGs) were conceived to include database statements and their dependency. Nevertheless, the use of syntax-based analysis in DOPDGs can be problematic and even lead to false information and redundant warnings. This paper suggests a semantics-based enhancement of dependency analysis for increased precision and efficiency. Through improved interpretation of dependencies, our approach reduces false positives and provides more reliable results. The effectiveness of this method is demonstrated through benchmark code evaluations.