Abstract Syntax Networks for Code Generation and Semantic Parsing
This addresses the challenge of generating structured outputs like code or semantic parses for developers and AI researchers, with competitive performance on multiple datasets without task-specific engineering.
The paper tackles the problem of mapping unstructured inputs to executable outputs in code generation and semantic parsing by introducing abstract syntax networks, achieving 79.2 BLEU and 22.7% exact match accuracy on the Hearthstone dataset, outperforming previous state-of-the-art results of 67.1 and 6.1%.
Tasks like code generation and semantic parsing require mapping unstructured (or partially structured) inputs to well-formed, executable outputs. We introduce abstract syntax networks, a modeling framework for these problems. The outputs are represented as abstract syntax trees (ASTs) and constructed by a decoder with a dynamically-determined modular structure paralleling the structure of the output tree. On the benchmark Hearthstone dataset for code generation, our model obtains 79.2 BLEU and 22.7% exact match accuracy, compared to previous state-of-the-art values of 67.1 and 6.1%. Furthermore, we perform competitively on the Atis, Jobs, and Geo semantic parsing datasets with no task-specific engineering.