Top-Down Tree Structured Text Generation
This addresses a bottleneck in natural language processing for generating structured text, though it appears incremental as it builds on existing tree-based methods.
The paper tackles the problem of generating long, complex sentences by modeling sentence generation as a top-down, breadth-first tree-generation task, which improves dependency handling and global planning compared to sequential models, with preliminary results showing effectiveness on generation and parsing tasks.
Text generation is a fundamental building block in natural language processing tasks. Existing sequential models performs autoregression directly over the text sequence and have difficulty generating long sentences of complex structures. This paper advocates a simple approach that treats sentence generation as a tree-generation task. By explicitly modelling syntactic structures in a constituent syntactic tree and performing top-down, breadth-first tree generation, our model fixes dependencies appropriately and performs implicit global planning. This is in contrast to transition-based depth-first generation process, which has difficulty dealing with incomplete texts when parsing and also does not incorporate future contexts in planning. Our preliminary results on two generation tasks and one parsing task demonstrate that this is an effective strategy.