Neural data-to-text generation: A comparison between pipeline and end-to-end architectures
This work addresses the problem of generating natural language from structured data for applications like automated reporting, though it is incremental as it compares existing architectures.
This study compared neural pipeline and end-to-end architectures for data-to-text generation from RDF triples, finding that pipeline models produced better texts and generalized better to unseen inputs.
Traditionally, most data-to-text applications have been designed using a modular pipeline architecture, in which non-linguistic input data is converted into natural language through several intermediate transformations. In contrast, recent neural models for data-to-text generation have been proposed as end-to-end approaches, where the non-linguistic input is rendered in natural language with much less explicit intermediate representations in-between. This study introduces a systematic comparison between neural pipeline and end-to-end data-to-text approaches for the generation of text from RDF triples. Both architectures were implemented making use of state-of-the art deep learning methods as the encoder-decoder Gated-Recurrent Units (GRU) and Transformer. Automatic and human evaluations together with a qualitative analysis suggest that having explicit intermediate steps in the generation process results in better texts than the ones generated by end-to-end approaches. Moreover, the pipeline models generalize better to unseen inputs. Data and code are publicly available.