A software company wants to protect its source code, including new source code created between scheduled indexing runs. Which detection method should the company use to meet this requirement?

A software company wants to protect its source code, including new source code created between scheduled indexing runs. Which detection method should the company use to meet this requirement?

• A. Vector Machine Learning (VML)
• B. Signature-Based Matching
• C. Heuristic Rules
• D. Rule-Based Scanning

Answer: (A)

Detecting new source code between indexing runs requires a learning-based approach that can generalize beyond fixed patterns. Vector Machine Learning analyzes features from code and builds a model that separates sensitive code from normal code. Once trained, it can flag fresh, unseen code that resembles the characteristics of sensitive patterns, even when there’s no exact signature to match. This makes it capable of catching code written after the last scheduled scan. In contrast, signature-based matching relies on exact or near-exact matches to known signatures and will miss novel code. Heuristic rules and rule-based scanning depend on predefined criteria and don’t adapt well to new coding styles or obfuscated patterns, requiring constant updates. Using a learning-based detector provides better generalization and timelier protection for source code.

Detecting new source code between indexing runs requires a learning-based approach that can generalize beyond fixed patterns. Vector Machine Learning analyzes features from code and builds a model that separates sensitive code from normal code. Once trained, it can flag fresh, unseen code that resembles the characteristics of sensitive patterns, even when there’s no exact signature to match. This makes it capable of catching code written after the last scheduled scan. In contrast, signature-based matching relies on exact or near-exact matches to known signatures and will miss novel code. Heuristic rules and rule-based scanning depend on predefined criteria and don’t adapt well to new coding styles or obfuscated patterns, requiring constant updates. Using a learning-based detector provides better generalization and timelier protection for source code.