We developed a general framework for the development of a symbolic (hand-written) feature-based lexicalised tree-adjoining grammar (FB-LTAG). We choose natural language generation, surface realisation in particular, to question the capabilities of the grammar in terms of both accuracy and robustness. Our framework combines an optimised surface realiser with efficient error mining techniques. While generating from a large data set provided by the Generation Challenge Surface Realisation task, we improve both accuracy and robustness of our grammar significantly.
|Title of host publication||1st International Data-to-Text Workshop|
|Number of pages||4|
|Publication status||Published - 6 Mar 2015|
|Event||1st International Data-to-Text Workshop - Edinburgh, United Kingdom|
Duration: 6 Mar 2015 → …
|Workshop||1st International Data-to-Text Workshop|
|Period||6/03/15 → …|