Accession Number:

AD1112162

Title:

PLINY: An End-To-End Framework for Big Code Analytics

Descriptive Note:

Technical Report,01 Sep 2014,31 Dec 2019

Corporate Author:

William Marsh Rice University Houston United States

Personal Author(s):

• Sarkar,Vivek

Report Date:

2020-10-01

Pagination or Media Count:

60.0

Abstract:

The PLINY project has addressed the grand challenge in the Defense Advanced Research Projects Agency program on Mining and Understanding Software Enclaves MUSE by advancing three key technical areas in the program Artifact Generators, Mining Engine, and Analytics. The PLINY project has introduced technologies that leverage the institutional knowledge embodied in the vast corpus of existing software to simplify the creation of new software. The main advances of the PLINY project have been in the following areas Code search Source Forager, Anomaly detection Salento, Test-based synthesis Splicer, SyPet, and Hunter, Evidence-based synthesis Bayou, and Scalable infrastructure for search and learning PlinyCompute. The PLINY technologies extend the reasoning capabilities of the programmer through the use of automated analysis and synthesis tools. The technologies developed in the PLINY project help address many of the productivity and cost challenges associated with software development today, and will also reshape the way people think about programming in the future.

Descriptors:

• bayesian networks
• neural networks
• computer programming
• computational science
• machine learning
• software development
• monte carlo method

Subject Categories:

• Computer Programming and Software
• Information Science

Distribution Statement:

APPROVED FOR PUBLIC RELEASE