In recent years, plagiarism in programming assignments and excessive code reuse in software development have become pressing concerns for both academia and the IT industry. Traditional text-matching tools often fail to identify copied programs when developers or students disguise them by altering variable names, rearranging code segments, or modifying formatting. This paper presents a web-based system designed to detect source code similarity by integrating structural analysis with Natural Language Processing (NLP) techniques. The proposed model evaluates code resemblance using multiple approaches, such as sequence-based (Difflib), vector-based (Cosine Similarity), token-based (Jaccard Index), and structure-based (Abstract Syntax Tree) methods. Among these, the AST-based technique stands out by focusing on the syntactic and structural patterns of code rather than its surface-level text. Experimental results demonstrate that this approach provides higher reliability and accuracy, especially when identifying refactored or slightly modified code segments.