Community Research and Development Information Service - CORDIS


PRIME Report Summary

Project ID: 615688
Funded under: FP7-IDEAS-ERC
Country: Israel

Mid-Term Report Summary - PRIME (Programming with Millions of Examples)

Project "Programming with Millions of Examples" (PRIME) combines program analysis, machine learning, and information retrieval techniques to make programming easier and faster by learning from all the code in the world.

In the first step of the project, we have developed several techniques for extracting semantic information from programs, and models that allow us to use this information for automatic code completion, program explanation, checking program similarity, and reverse engineering. The techniques that we have developed make programmers more productive and can assist in understanding software and finding security vulnerabilities.

* Code Acquisition: We have developed techniques for data extraction and crawling to facilitate efficient and precise code acquisition. We have also developed techniques for correlating code with natural language descriptions, and techniques for extracting code from challenging containers, such as tutorials sites and tutorial videos.

* Code Analysis: We have developed new representations of code that can be efficiently and statical extracted from programs (both at source level, and at binary level).

* Learning from Code: We have developed models for representing (statistical) information about code, and techniques for using these models for code completion, prediction of program properties, etc.

* Applications: We have used the above techniques for challenging applications such as code completion, prediction of program properties (variable names, and others), similarity between programs and between stripped binaries.

Reported by

Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top