From Partial to Strictly Incremental Constituent Parsing
This addresses the need for more reliable incremental parsing in natural language processing applications.
The researchers tackled the problem of building truly incremental constituent parsers that can generate trees using only prefix representations, achieving strong performance across multiple languages by enforcing strict left-to-right processing in both encoder and decoder components.
We study incremental constituent parsers to assess their capacity to output trees based on prefix representations alone. Guided by strictly left-to-right generative language models and tree-decoding modules, we build parsers that adhere to a strong definition of incrementality across languages. This builds upon work that asserted incrementality, but that mostly only enforced it on either the encoder or the decoder. Finally, we conduct an analysis against non-incremental and partially incremental models.