A General Path-Based Representation for Predicting Program Properties
Predicting program properties such as names or expression types has a wide range of applications. It can ease the task of programming, and increase programmer productivity. A major challenge when learning from programs is \emph{how to represent programs in a way that facilitates effective learning}.
We present ageneral path-based representation for learning from programs. Our representation is purely syntactic and extracted automatically. The main idea is to represent a program using paths in its abstract syntax tree (AST). This allows a learning model to leverage the structured nature of code rather than treating it as a flat sequence of tokens.
We show that this representation is general and can:
(i) cover different prediction tasks,
(ii) drive different learning algorithms (for both generative and discriminative models), and
(iii) work across different programming languages.
We evaluate our approach on the tasks of predicting variable names, method names, and full types. We use our representation to drive both CRF-based and word2vec-based learning, for programs of four languages: JavaScript, Java, Python and C#. Our evaluation shows that our approach obtains better results than task-specific handcrafted representations across different tasks and programming languages.
Thu 21 JunDisplayed time zone: Eastern Time (US & Canada) change
14:00 - 15:40 | Synthesis and LearningPLDI Research Papers at Grand Ballroom CD Chair(s): Xin Zhang Massachusetts Institute of Technology, USA | ||
14:00 25mTalk | A General Path-Based Representation for Predicting Program Properties PLDI Research Papers Uri Alon Technion, Meital Zilberstein Technion, Omer Levy University of Washington, USA, Eran Yahav Technion Media Attached | ||
14:25 25mTalk | Program Synthesis using Conflict-Driven Learning PLDI Research Papers Yu Feng University of Texas at Austin, USA, Ruben Martins Carnegie Mellon University, Osbert Bastani Stanford University, Işıl Dillig UT Austin Media Attached | ||
14:50 25mTalk | Accelerating Search-Based Program Synthesis using Learned Probabilistic Models PLDI Research Papers Woosuk Lee University of Pennsylvania, USA, Kihong Heo University of Pennsylvania, USA, Rajeev Alur University of Pennsylvania, Mayur Naik University of Pennsylvania Media Attached | ||
15:15 25mTalk | Inferring Crypto API Rules from Code Changes PLDI Research Papers Rumen Atanasov Paletov , Petar Tsankov ETH Zurich, Veselin Raychev ETH Zurich, Martin Vechev ETH Zürich Media Attached | ||