Antecedent Predictions Are Dominant for Tree-Based Code Generation

Code generation focuses on the automatic conversion of natural language (NL) utterances into code snippets. The sequence-to-tree (Seq2Tree) methods, e.g., TRANX, are proposed for code generation, with the guarantee of the compilability of the generated code, which generate the subsequent Abstract Syntax Tree (AST) node relying on antecedent predictions of AST nodes. Existing Seq2Tree methods tend to treat both antecedent predictions and subsequent predictions equally. However, under the AST constraints, it is difficult for Seq2Tree models to produce the correct subsequent prediction based on incorrect antecedent predictions. Thus, antecedent predictions ought to receive more attention than subsequent predictions. To this end, in this paper, we propose an effective method, named APTRANX (Antecedent Prioritized TRANX), on the basis of TRANX. APTRANX contains an Antecedent Prioritized (AP) Loss, which helps the model attach importance to antecedent predictions by exploiting the position information of the generated AST nodes. With better antecedent predictions and accompanying subsequent predictions, APTRANX significantly improves the performance. We conduct extensive experiments on several benchmark datasets, and the experimental results demonstrate the superiority and generality of our proposed method compared with the state-of-the-art methods.