SLIDE 1
Mining Source Code Repositories at Massive Scale using Language Modeling
Miltos Allamanis, Charles Sutton
m.allamanis@ed.ac.uk csutton@inf.ed.ac.uk
University of Edinburgh
Supported by:
Mining Source Code Repositories at Massive Scale using Language - - PowerPoint PPT Presentation
Mining Source Code Repositories at Massive Scale using Language - - PowerPoint PPT Presentation
Mining Source Code Repositories at Massive Scale using Language Modeling Miltos Allamanis, Charles Sutton m.allamanis@ed.ac.uk csutton@inf.ed.ac.uk University of Edinburgh Supported by: Polyglot programmers Multitude of APIs &
SLIDE 2
SLIDE 3
Why Language Models?
- Statistical models
- Learn from data
- Abundance of code
available online
- Non-language specific
method
[Hindle et al., ICSE 2012]
SLIDE 4
n-gram Language Models
public void execute(Runnable task) { if (task == null) throw new NullPointerException(); ForkJoinTask<?> job; if (task instanceof ForkJoinTask<?>) // avoid re-wrap job = (ForkJoinTask<?>) task; else job = new ForkJoinTask.AdaptedRunnableAction (task); externalPush(job); }
SLIDE 5
n-gram Language Models
public void execute(Runnable task) { if (task == null) throw new NullPointerException(); ForkJoinTask<?> job; if (task instanceof ForkJoinTask<?>) // avoid re-wrap job = (ForkJoinTask<?>) task; else job = new ForkJoinTask.AdaptedRunnableAction (task); externalPush(job); }
SLIDE 6
n-gram Language Models
public void execute(Runnable task) { if (task == null) throw new NullPointerException(); ForkJoinTask<?> job; if (task instanceof ForkJoinTask<?>) // avoid re-wrap job = (ForkJoinTask<?>) task; else job = new ForkJoinTask.AdaptedRunnableAction (task); externalPush(job); }
SLIDE 7
n-gram Language Models
public void execute(Runnable task) { if (task == null) throw new NullPointerException(); ForkJoinTask<?> job; if (task instanceof ForkJoinTask<?>) // avoid re-wrap job = (ForkJoinTask<?>) task; else job = new ForkJoinTask.AdaptedRunnableAction (task); externalPush(job); }
SLIDE 8
n-gram Language Models
Predictability Measures n-gram Log Probability (NGLP) Cross-Entropy (H)
SLIDE 9
The Java GitHub Corpus
Java projects >1 fork Deduplication through git commit SHAs URL: http://groups.inf.ed.ac.uk/cup/javaGithub/
SLIDE 10
Language Models of Code
SLIDE 11
Learning about identifiers
SLIDE 12
Learning about identifiers
API calls are predictable
SLIDE 13
n-gram log probability (NGLP) as a complexity metric
NGLP is Data-Driven
An n-gram is more complex if it is more rare
SLIDE 14
Complexity trade-offs
from elasticsearch
SLIDE 15
vs
from elasticsearch
SLIDE 16
Evaluate domain specificity of code Larger IIM, more domain specific identifiers Use to evaluate code reusability
Identifier Information Metric (IIM)
Hfull - Hcollapsed
ContinuationPending.java 5.2 FastDtoa.java 5.0 PrivateAccessClass.java 4.7 JSSetter.java 1.0 GeneratedClassLoader. java 1.1 UintMap.java 1.2
SLIDE 17
Contributions
- GitHub Java Corpus
- New gigatoken language models
- API calls are predictable
- Data-driven code complexity metrics
- Metric of domain-specificity
SLIDE 18
Mining Source Code Repositories at Massive Scale using Language Modeling
Miltos Allamanis, Charles Sutton
m.allamanis@ed.ac.uk csutton@inf.ed.ac.uk
University of Edinburgh
Supported by:
SLIDE 19
n-gram Language Models
SLIDE 20
Language Models - Metrics
Log Probability (NGLP) Cross Entropy (H)
SLIDE 21
Learning about identifiers
SLIDE 22
Learning about identifiers
Method and Type identifiers are equally hard, irrespectively of the amount of data.