Test Suites as a Source of Training Data for Static Analysis Classifiers

Abstract

We developed a novel method that uses test suites to automatically generate labeled data for static analysis classifiers. We implemented the method in a software system. In a case study, we generated a large quantity of labeled data for many CWE, using the Juliet C/C++ v 1.2 test suite. With that, we created 4 types of classifiers and tested them on holdout data. We tested speculative mapping and devised an effort-efficient part-automated method to map static analysis tools to test suite taxonomies.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 21, 2021
Accession Number
AD1132177

Entities

People

  • Lori A. Flynn

Organizations

  • Carnegie Mellon University

Tags

DTIC Thesaurus Topics

  • Accuracy
  • Analyzers
  • Artificial Intelligence
  • Automation
  • Case Studies
  • Classification
  • Copyrights
  • Department Of Defense
  • Engineering
  • Governments
  • Guarantees
  • Learning
  • Line Defects
  • Machine Learning
  • Materials
  • Metadata
  • Software Development
  • Software Testing
  • Taxonomy
  • Training
  • Universities

Fields of Study

  • Computer science
  • Engineering

Readers

  • Aerospace Test and Evaluation
  • Neural Network Machine Learning.
  • Software Engineering.